Evmarbella's Web Log

FSOSS '09 - Poster Design & Business Card [1.1]

2009-06-13T13:39:00.012-04:00

Creatin' new poster for FSOSS '09 [first draft 001]

Wanted to keep the old style of the landscape design.
Attempted to keep all the information relatively in the same place for familiarity.
[information/image below is subject to change]

Poster 11x17

Poster Variation [001]

Poster Variation [002]

Business Card 3.5x2

Academic source code dust-up symptom of CS education ills

2009-06-12T18:10:00.003-04:00

http://digg.com/d1tfeB

mozilla firefox 3.5 and libnotify

2009-05-04T22:07:00.006-04:00

building mozilla firefox on Ubuntu 9.04
a slight update.

You want up update your build dependencies to 3.5
sudo apt-get build-dep firefox-3.5

NEW packages used:
libstartup-notification0-dev
xulrunner-1.9.1
xulrunner-1.9.1-dev

libnotify issues...if you run into these
you will need to install libnotify-dev not libnotify-bin.

sudo apt-get install libnotify-dev

create a .mozconfig
ref: https://developer.mozilla.org/en/Configuring_Build_Options

then it builds...in 'mozilla-central'
make -f client.mk build

~/mozilla/mozilla-central/obj*/dist/bin/firefox -Profilemanager -no-remote

Mochitests on Fennec [bookmarks][tabs][1.oh]

2009-04-23T16:35:00.003-04:00

So got a r+ on tabs tests. Which is dandy!
https://bugzilla.mozilla.org/show_bug.cgi?id=485855
Got the hang of things on the bookmark container.
https://bugzilla.mozilla.org/show_bug.cgi?id=486490#c5
did a forced button click save but it didn't save the url.

A set of test coverage/bugs was wiki page recently created
https://wiki.mozilla.org/Mobile/Fennec_TestDev
So things are just getting started...it's exciting.

ref:
Tabs
https://bugzilla.mozilla.org/attachment.cgi?id=373599&action=diff&context=patch&collapsed=&headers=1&format=raw

Bookmarks
https://bugzilla.mozilla.org/attachment.cgi?id=373711&action=diff&context=patch&collapsed=&headers=1&format=raw

Mochitests on Fennec [bookmarks][tabs][update]

2009-04-20T13:35:00.002-04:00

There's a wiki page!! huzzah.
https://wiki.mozilla.org/Mobile/Fennec_TestDev

funtimes...

Mochitests on Fennec [tabs][bookmarks][0.9]

2009-04-12T20:12:00.003-04:00

So an updated patch for Tab testing on Fennec.
https://bugzilla.mozilla.org/show_bug.cgi?id=485855
able to test for tab titles and switching between tabs. hurah.

Bookmarks is a bit more challenging...(want to know how this all works)
note: thought it was just getElementById...

How bookmarks work(on Fennec)...the bookmark panel can be accessed
by window.BrowserUI.doCommand("cmd_star");
which calls on this window
http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser.xul#338
the "editor" portion is here
http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser.xul#341
the editors place-item binding is on
http://mxr.mozilla.org/mobile-browser/source/chrome/content/bindings.xml#457
and the textbox is
http://mxr.mozilla.org/mobile-browser/source/chrome/content/bindings.xml#467

so one can use document.getAnonymousElementByAttribute to get the value...
currently unable to get it to giv'r...

ref: https://bugzilla.mozilla.org/show_bug.cgi?id=486490

Mochitests on Fennec [tabs][bookmarks]

2009-04-03T06:53:00.005-04:00

A couple of things to work on this week.
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=485855#c4
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=486490#c2

updated/added more tab tests
and working on bookmark tests...

Mochitests on Fennec (browser chrome bookmarks...patch)

2009-04-02T08:55:00.002-04:00

Mochitests on Fennec (not on device)
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=486490
Testing to see if a bookmark has been added on the bookmark list.

Bookmark service will probably be useful for this...maybe Bookmark List
ref: http://mxr.mozilla.org/mozilla-central/source/toolkit/components/places/tests/queries/head_queries.js#131
ref: http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser-ui.js#641

not sure yet how Fennec deals with listing bookmarks.

Mochitests on Fennec (browser chrome bookmarks...getting started)

2009-04-01T17:49:00.005-04:00

So started creating the test got the add to work but not
sure how to check to see if it's been added to the list.

ref: http://mozilla.pastebin.com/f64123d5b

want to test for
-add (works)
-list
-remove

slowly getting use to the browser-ui and browser.
ref: http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser-ui.js
ref: http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser.js
ref: http://mxr.mozilla.org/mozilla-central/source/toolkit/components/places/tests/queries/head_queries.js#95

reading up on bookmarks on Firefox
ref: https://developer.mozilla.org/en/Code_snippets/Bookmarks

Mochitests on Fennec (browser chrome tabs) [0.8]

2009-03-29T23:04:00.003-04:00

Filed a bug that has a patch that tests for tab creation and removal on Fennec.
https://bugzilla.mozilla.org/show_bug.cgi?id=485855

It's been very interesting looking around and trying to
figure out how to write the tests.

Haven't been able to find documentation on how to write
the tests and been going on examples on the mxr...
https://developer.mozilla.org/en/Mochitest

Theres like assertions like
is(function, expected result, message)
isnot(function, expected result, message)
ok(function, message)
ref: http://mxr.mozilla.org/mozilla-central/source/browser/fuel/test/browser_Browser.js

In order to run the tests on Fennec one must actually go to the

$objdir/mobile/tests/testing/mochitests

then run this command

python runtests.py --appname=$objdir/mobile/dist/bin/fennec --xre-path=$objdir/xulrunner/dist/bin --browser-chrome

note: the --xre-path, --appname is your path to fennec
note: it is similar to how "To test on older branches"

Mochitests on Fennec (more tabs)

2009-03-29T15:08:00.004-04:00

Updated the Xulrunner and rebuilt...

Mochitests for adding and closing Tabs updated
http://mozilla.pastebin.com/f4301ff30

So I think I'm getting the hang of it.
And the Tests seems to be reacting as I expected.
exciting...

Mochitests on Fennec (more)

2009-03-28T20:46:00.002-04:00

Starting to write mochitests for Fennec
ref: https://developer.mozilla.org/en/Mochitest

Writing Tests for Tabs
http://mozilla.pastebin.com/f3f34df14
opening
http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser.js#286
and
closing of tabs
http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser.js#301

Currently it's giving me this output.
http://mozilla.pastebin.com/f5146872b

Not sure if this is the test or something on Fennec...
going to find out more.

Building and Testing Documentation [0.6]

2009-03-20T07:35:00.003-04:00

Created a wiki page with instructions for Building and Testing on Fennec on a Nokia N810
http://zenit.senecac.on.ca/wiki/index.php/Fennec_on_a_Nokia_N810

Bug 468214 - Add mochitest support for Fennec update

2009-03-18T19:15:00.002-04:00

Test was failing because it was testing
for the about:firstrun it's suppose to be about:blank

resubmitted with the changes
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=468214#c8

gdb on Fennec (nokia) revisited

2009-03-18T09:45:00.003-04:00

ref: http://zenit.senecac.on.ca/wiki/index.php/Fennec_on_a_Nokia_N810#Debugging_on_Fennec_on_a_Nokia_N810

note: still doesn't work. :(

what we can do so far:
- you can install it on the device.
- you can package it with the symbols

what doesn't work:
- the -g flag
ref: https://developer.mozilla.org/en/Debugging_Mozilla_on_Linux_FAQ#How_do_I_debug_Mozilla_on_Linux.3f
- running gdb ./fennec
(reboots the device if you ssh into it and if you run it on the device itself the process gets killed)

const to let [0.7]

2009-03-17T14:53:00.007-04:00

ref: https://developer.mozilla.org/en/New_in_JavaScript_1.7#Block_scope_with_let
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=468214

So far when running --browser-chrome tests in Fennec the harness doesn't load.
and javascript error used to occur

ex. JavaScript error: chrome://mochikit/content/browser-test.js, line 5: redeclaration of const Cc

Had to change 'let' instead of 'const'...(has been changed)
ref: http://mxr.mozilla.org/mobile-browser/source/chrome/content/browser.js#44

note: "The let statement provides local scoping for variables" see https://developer.mozilla.org/en/New_in_JavaScript_1.7#Block_scope_with_let for more details

with those changes made the tests loads...hurah.

output of the test
http://mozilla.pastebin.com/f78645dc1

--browser-chrome test harness

2009-03-11T10:13:00.007-04:00

It is not loading properly.
in runtest.py...

note: After making necessary changes to accommodate
browser chrome tests in fennec...moving files and changing the test.
patch

runtest.py
creates a chrome manifest
and enables overlays for testing
ref: https://developer.mozilla.org/en/Chrome_Registration#overlay
that is the function of the browser chrome harness

Currently we get a blank page...not sure why that is yet.

note: started to look into how to debug python
http://www.ferg.org/papers/debugging_in_python.html

Browser-Chrome Mochitests on Fennec rev.1

2009-03-08T22:03:00.002-04:00

Getting started....

Managed to move the files in the proper folders and
created a browser-chrome test which uses .js files instead of
chrome tests .xul files

However, --browser-chrome flag for running the tests
doesn't work
note: in order to get the test to work we must specify
--xre-path on linux
note: just trying to get the --browser-chrome to work
on linux first

This current patch does that.
http://matrix.senecac.on.ca/~evmarbella/mochifenn.patch

Will look into why --browser-chrome doesn't work.

Fennec Mochitest using the original runtest.py

2009-03-06T16:02:00.003-05:00

After building with the patch I replaced the runtests.py that the patch created
with the standard runtests.py from mozilla-central.

http://mxr.mozilla.org/mozilla-central/source/testing/mochitest/runtests.py.in

Ran into an error with libxpcom.so. not sure if this is fennec related or something else.

seneca@seneca-desktop:~/mozilla/mochifenn/mobilebase/mobile/_tests/testing/mochitest$ python runtests.py --appname=/home/seneca/mozilla/mochifenn/mobilebase/mobile/dist/bin/fennec --browser-chrome --autorun
/home/seneca/mozilla/mochifenn/mobilebase/xulrunner/dist/bin/xpcshell: error while loading shared libraries: libxpcom.so: cannot open shared object file: No such file or directory
Server pid: 1572

note: there is a lot of material to understand.

ref: https://developer.mozilla.org/en/Browser_chrome_tests

runtests diff

2009-03-04T11:54:00.001-05:00

An attempt to look at the differences between the patches
runtests.py and the one on mozilla-central

http://mozilla.pastebin.com/f3688b594

Mochitests on Fennec

2009-03-02T10:23:00.007-05:00

ref: https://bugzilla.mozilla.org/show_bug.cgi?id=468214
Currently browser chrome is not in the runtest.py as one of the flags

so we add it in
ref: http://mxr.mozilla.org/mozilla-central/source/testing/mochitest/runtests.py.in

built it and ran it in the
cd $objdir/mobile/_tests/testing/mochitests

python runtests.py --appname=$objdir/mobile/dist/bin/fennec --browser-chrome --autorun

The patch currently works on the desktop...
however, not sure how to move things over to the device yet.

Build Env. broke and scratcbox is down

2009-02-27T10:23:00.003-05:00

Don't understand why
mmap: Permission Denied even when running as fakeroot.
then even after installing by running

fakeroot apt-get install libgtk2.0-dev libxaw7-dev libfreetype6-dev libxi-dev libhildonmime-dev libosso-dev libidl-dev liblocation-dev libasound2-dev libhildon1-dev libxdamage-dev libxcomposite-dev

from
https://wiki.mozilla.org/Mobile/Build/cs2007q3

It told me it wasn't installed...in any case going to try to set things up again.
but scratchbox.org is down as of 10:27AM Fri, 27/09...lovely.

1 patch and a broken build...

2009-02-22T23:23:00.003-05:00

So after a better understanding on the challenges
for testing idle service

Doug provided a solution on how to test for it
in a form of a patch that changes what the browser
uses to test for idle into the screen dimmin
https://bugzilla.mozilla.org/show_bug.cgi?id=472548#c7

Attempted to run the tests on the build...got an error.
http://mozilla.pastebin.com/fb16b08d
(this is probably a problem)
the build worked on it's own...so I thought it
might have been xulrunner.

So downloaded/updated to latest source.

Applied the patch(es) on a fresh source.
And got a different build Error.
http://mozilla.pastebin.com/f50c7c938
something in MaemoLocationProvider.cpp
causing issues...and it is not letting me build.

This was apparently a bug.
Will get to the bottom of this.

My scratchbox broke...again

2009-02-22T19:14:00.003-05:00

It keeps looking for the libIDL pkconfig path
and I export the path in $PKG_CONFIG_PATH
and it's not givin'r...

installed scratchbox again.
and tried to build with the patches and got
an error in MaemoLocationProvider.cpp

http://mozilla.pastebin.com/f597da0e9

gonna try again...

In light of this new information...update on Bug 472548

2009-02-21T17:17:00.003-05:00

Found out that it is a challenge
to test for idle services on mobile
devices and that it too is being worked
on Windows Mobile

https://bugzilla.mozilla.org/show_bug.cgi?id=475361
it uses (soft keypad) events for idle time

But there are some insights to the problem
one by Doug
https://bugzilla.mozilla.org/show_bug.cgi?id=472548#c8
and another by Gijs
https://bugzilla.mozilla.org/show_bug.cgi?id=472548#c9

Doug's insight involves using a different
service to determine if the system is idle
instead of osso_hw_set_event_cb
it uses the OSSO_DISPLAY_ON.

Gij's insight involves 'fooling' the OSSO
component into being idle.
(not sure if what we did before applies...
waiting for more information to come in)

testing the build but it's not workin' with
the chrome test...may need to rebuild by
applying the patch to the source and building.

Is it really going idle?...I mean really?

2009-02-18T17:59:00.005-05:00

It has been theorised that idle service might not
be actually going idle...since the idle time is so short

http://mxr.mozilla.org/mozilla-central/source/widget/tests/test_bug343416.xul#112

5 seconds to be exact.

Tried to adjust the setTimeout to another number

112 setTimeout(testIdleTime, 600000); (which would be 10 mins)
but that didn't want to giv'r

so came up with another way to count
I added this at after line 112.
http://mozilla.pastebin.com/f3cb64477 (a somewhat counter)

So I've been spending so much time waiting...

and after thinking it hung up several times.
We ended up using 10000 as the number
for idle. It was good the test was "idling"
in the loop.

But we kept getting negative numbers
right after the loop and the test queries
the interface...and then tries to access
the GetIdleTime function inside nsIdleOSSO.cpp.

funtimes...
note: the loop occurs between line 6 and 7
http://mozilla.pastebin.com/ff16a464

Then just for kicks ran the test on the Linux build of Fennec.
with the loop the test fails on Linux

setTimeout(testIdleTime, 60000); The test fails
setTimeout(testIdleTime, 10000); The test passes

? whats goin' on har ?

another interesting thing
the flag --close-when-done
didn't work during the chrome test
on the Linux build of Fennec.

How does the nsIdleOSSO.cpp work?

2009-02-18T09:06:00.004-05:00

I just received answers to my emails and they helped
clear things up with how I thought nsiIdleOSSO works.

http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceOSSO.cpp

So this was from the email:
The "I" here is Gijs Kruitbosch

1) the system will tell the app when it is idle, and expects the app
to reduce its activity (
http://maemo.org/api_refs/4.0/libosso/structosso__hw__state__t.html ,
http://maemo.org/api_refs/4.0/libosso/group__Devstate.html#gdbbda331888e984d67efe4b231e51791).
2) the callback that's registered (
http://mxr.mozilla.org/mozilla-central/source/toolkit/xre/nsNativeAppSupportUnix.cpp#306
, http://mxr.mozilla.org/mozilla-central/source/toolkit/xre/nsNativeAppSupportUnix.cpp#215
) sends a notification to the observerservice. (note: this is how I
found this - I looked in MXR for "system-idle", which is the
notification topic)
3) the OSSO idle service picks up on that notification, sets the time
it received it, and calculates idle time based on that. That's what it
uses for the result of GetIdleTime().

There's more things to come.
sidenote: so a fennec build on Linus uses nsIdleServiceGTK.cpp which
instead of calling on the hw state uses the xscreensaver to calculate idle
time...I found that interesting.

A working test to compare to

2009-02-17T21:45:00.001-05:00

Having a working and non-working
chrome test does help discern the differences .

But why does the ARM use an observe and the Linux doesn't?
I take it they use 2 different criteria to detect if a system is idle?

http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceGTK.cpp

http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceOSSO.cpp

Also is there a way to find out if these observe(s)
get registered at app-start up in the Category
Manager in Fennec?

Running Fennec Chrome tests on Linux...and stale Valentines Day Candy

2009-02-15T20:12:00.005-05:00

So the chrome tests didn't work because we didn't
have the paths leading to utility-path, certificate-path
and xre-path when we try to run the chrome test
(test systems change...)

To run the fennec chrome test on linux

python runtests.py --appname=../../../../mobile/dist/bin/fennec --chrome --utility-path=~/mozilla/mobilebase/xulrunner/dist/bin --certificate-path=~/mozilla/mobilebase/xulrunner/dist/bin --xre-path=~/mozilla/mobilebase/mobile/dist/bin/xulrunner --autorun --close-when-done

It was fun trying to find where these folders/paths were.
Then had to replace the chrome test with the one with the
dump() on it.

Then ran the test again it gave me this output

http://mozilla.pastebin.com/f4a19d9bb (on the Ubuntu)

compared to the one on the device it is different...

http://mozilla.pastebin.com/f5825f5ee (on the N810)

note: DEBUG OSSO means its a printf() on the nsIdleServiceOSSO.cpp

It was actually setting idleTime and newIdleTime
compared to the test on the device which stayed
at 0...oh and the chrome tests pass.

Runing Chome test on Fennec Linux build rev1

2009-02-13T07:56:00.004-05:00

So after getting a fresh clone of the source and rebuilding it
without the patches necessary for the n810.

Attemped to run the chrome tests and ran into this.

seneca@seneca-desktop:~/mozilla/mobilebase/xulrunner/_tests/testing/mochitest$ python runtests.py --appname=../../../../mobile/dist/bin/fennec --chrome
/home/seneca/mozilla/mobilebase/xulrunner/dist/bin/xpcshell: error while loading shared libraries: libxpcom.so: cannot open shared object file: No such file or directory
Server pid: 27206
Timed out while waiting for server startup.

tried to look for the error on the mxr and unable to find it.

Then attempted to apply the patches from before(which had no issues)
and ran into

seneca@seneca-desktop:~/mozilla/mozilla-central$ hg qnew gredir
seneca@seneca-desktop:~/mozilla/mozilla-central$ patch -p1 < Bug470971.txt
patching file config/autoconf.mk.in
Reversed (or previously applied) patch detected! Assume -R? [n] y
Hunk #1 succeeded at 181 (offset 2 lines).
patching file configure.in
Reversed (or previously applied) patch detected! Assume -R? [n] y
Hunk #1 succeeded at 7296 (offset 75 lines).
patching file js/src/xpconnect/shell/Makefile.in
Hunk #1 FAILED at 45.
1 out of 1 hunk FAILED -- saving rejects to file js/src/xpconnect/shell/Makefile.in.rej
patching file js/src/xpconnect/shell/xpcshell.cpp
Reversed (or previously applied) patch detected! Assume -R? [n] y
Hunk #3 succeeded at 914 (offset 5 lines).
Hunk #4 succeeded at 1543 (offset 5 lines).
Hunk #5 succeeded at 1711 (offset 5 lines).

Chose yes for it and some failed and some passed.
however the build fails...Gonna ask around to see what's going on with
this...

Running the Test on Linux

2009-02-12T22:47:00.006-05:00

It was proposed for me to run the tests on Linux

https://bugzilla.mozilla.org/show_bug.cgi?id=472548#c5

This should be pretty straight forward ...
go into xulrunner folder
isolating the test (widget)
then running the test
ref: http://zadkielm.blogspot.com/2009/01/fennec-and-chrome-testing.html

however I'm running into

error while loading shared libraries: libxpcom.so

Not sure what is causing this.
Maybe we need to apply the patches...
ref: http://zadkielm.blogspot.com/2009/01/so-why-cant-we-run-tests.html

going to clone the source again and giv'r.

when applying the patches I got this

Reversed (or previously applied) patch detected! Assume -R? [n]

and I opted for y.
y for the patches caused build errors.

while waiting for it to build
note: the might be using the
http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceGTK.cpp
and not the
http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceOSSO.cpp

Meanwhile on IRC and Bugzilla...observe

2009-02-10T22:53:00.006-05:00

Got a text message today from Armen to go on IRC...

As I was putting posters up for a school event.
A screening of TKARONTO at Seneca. see the posters for details...
http://www.tkaronto.net/

As soon as I was finished I logged on.
Dave(humph) on IRC recommended a solution as to why the observer not being called.

dave:
[13:57] I wonder if you need to register your stuff with app-startup via the Category Manager so you get created in time to add those two observers
[13:58] i.e., I wonder if your init is happening after your chance to register has gone by

ref:
http://mxr.mozilla.org/mozilla-central/source/toolkit/components/alerts/src/mac/nsAlertsServiceModule.cpp#63

https://bugzilla.mozilla.org/attachment.cgi?id=341439&action=diff

But as I was walking around later that day replacing the posters I just put up...
I got to thinking would the Linux version and windows version require the same registration?

Then Joel commented on the bug with an idea
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=472548#c5
to run the chrome test the linux build of fennec.

I suppose we'll have to copy over the test via bluetooth...I would have never thought
of me using bluetooth to "pair" devices to transfer files. It feels fancy...

I'll probably get around looking into this sometime this week...

When printf and dump are all you have...[0.5]

2009-02-09T19:53:00.005-05:00

So when was looking for some insight on how to start debugging on the n810
I was under the impression that there are a set of tools that would make things easier.
This was not the case...I was told to printf and dump things.

At first I was a bit lost on where to put these things but when you get started you just
want to put them everywhere just to see things...

ref:
http://zadkielm.blogspot.com/2009/02/observerservice-and-bug.html

note: the best part is that if you put DEBUG on all of them you can just grep it
and they assemble on their own...that was so spiffy.

Right now I am trying to better understand how the nsIdleServiceOSSO works
along with test on Bug 472548.

It's not calling the observe which sets our mIdle to calculate I added another
comment on the bug in reference to it...someone will explain it to me.

https://bugzilla.mozilla.org/show_bug.cgi?id=472548#c4

Also this morning on the road to figuring out how we can get gdb to work
on the nokia n810 to debug...we found out that the package strips the symbols.

file fennec

Found out that you can't just copy the symbols over...and the person on irc
then mentions that nokia is going to kill the n810 series. :( good news all around.

In any case I'm thinking of just copying the whole build over to the device but that might
take a while to copy over to the device itself...

ObserverService and the Bug

2009-02-08T23:00:00.005-05:00

We are trying to understand what these observer services
http://www.xulplanet.com/references/xpcomref/ifaces/nsIObserverService.html
https://developer.mozilla.org/en/NsIObserverService

http://mxr.mozilla.org/seamonkey/source/xpcom/ds/nsObserverService.cpp#124
https://developer.mozilla.org/en/nsObserverService

updated DEBUG output on the test_...xul file and the nsIdleServiceOSSO

DEBUG Test nsIIdleService: nsIIdleService
DEBUG Test nsIISCID: @mozilla.org/widget/idleservice;1
DEBUG OSSO inside constructor nsIdleServiceOSSO()
DEBUG Test idleService: [xpconnect wrapped nsIIdleService @ 0x45ac27c0 (native @ 0x4650d130)]
DEBUG OSSO aTimeDiff: 0
DEBUG Test idleTime: 0
DEBUG Test QueryInterface iid: {00000000-0000-0000-c000-000000000046}
DEBUG Test nsISupports: true
DEBUG Test nsIObserver: false
DEBUG OSSO aTimeDiff: -1094995628
DEBUG Test idleService.addIdleObserver: undefined
DEBUG Test QueryInterface iid: {00000000-0000-0000-c000-000000000046}
DEBUG Test nsISupports: true
DEBUG Test nsIObserver: false
DEBUG OSSO aTimeDiff: -1094995628
DEBUG Test idleService.IdleObserver second time: undefined
DEBUG Test QueryInterface iid: {00000000-0000-0000-c000-000000000046}
DEBUG Test nsISupports: true
DEBUG Test nsIObserver: false
DEBUG Test idleService.removeIdleObserver: undefined
DEBUG OSSO aTimeDiff: 0
DEBUG OSSO aTimeDiff: 0
DEBUG Test newIdleTime: 0
DEBUG Test var timeDiff: 5016
DEBUG OSSO aTimeDiff: -636291451
DEBUG OSSO aTimeDiff: 795364314
DEBUG Test QueryInterface iid: {00000000-0000-0000-c000-000000000046}
DEBUG Test nsISupports: true
DEBUG Test nsIObserver: false
DEBUG OSSO inside destructor ~nsIdleServiceOSSO

nsIdleServiceOSSO.cpp w/ DEBUG http://mozilla.pastebin.com/f19d1ae71
test_bug343416.xul w/ DEBUG http://mozilla.pastebin.com/f7321de69

Dyndns comes to the rescue

2009-02-08T01:31:00.004-05:00

So we have our device setup and currently
if we are at school we are unable to work
on the bug since Senenet only allows 1
wireless connection at a time.

However, if we setup a dyndns account
at
https://www.dyndns.com/
we can keep our device n810 plugged
in at all times at home and
ssh to it using via dyndns.

fun
so theoretically (I only tried from home...hopefully the
school doesn't block the ssh ports)

ssh root@your.dyndns.address.com
password

tada...

Installing GDB on the N810 to work with Fennec rev.1

2009-02-08T00:00:00.003-05:00

This is the most exciting thing ever once we figure out how it works
debugging would be so much easier on there.

installing gdb
ref: http://maemo.org/development/tools/#apt-example

on your device

To activate the tools repository with the Application manager, create a new catalogue like this:

Start the Settings > Application manager
From its menu bar select Tools > Application catalogue...
Press the New button
Enter the following:

Catalogue name:	diablo tools
Web address:	http://repository.maemo.org
Distribution:	diablo/tools
Components:	free non-free
Disabled:	leave unchecked

Click OK

now we should be able to ssh into the device
and run

sudo apt-get install gdb

now since we have been packaging and copying our Fennec
builds to the device they do not include the symbols necessary that gdb needs.
ref: http://zadkielm.blogspot.com/2009/01/bits-of-fennec-for-device.html

we need to manually copy over the symbols...not sure where they are located yet
or how to set them up. another option that was suggested was remote gdb...no clue
what that is.

so 'til we figure that out it's back to dump() and printf().

note: on the bug472548 the Observer in
ref: http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceOSSO.cpp#70
doesn't seem to get called so mIdle is not set in line
ref: http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceOSSO.cpp#61
during the test
ref: http://mxr.mozilla.org/mozilla-central/source/widget/tests/test_bug343416.xul

Debuggin Fennec on Maemo(N810) - Bug 472548

2009-02-07T16:54:00.007-05:00

Got the device up and running and I'm able to debug the
xul tests. The N810 is pretty spiffy...I finally got to ssh to the
silly thing and got it to take some pictures at the FLOSSLC

When my build works again...getting things just to the point
of just working with chrome tests and what not is proving to be a
challenge. To top things off everything is slower on mobile.

I really want lay down some printfs and see what's going on.
nsIdleServiceOSSO.cpp
But I first want to make sure I'm doing things correct...it's hard to tell.

I was doing some dump() in the test file which was helpful
in finding out what Bug 472548 is doing.

Joel(jmaher) was very helpful getting me started...along with bc
in qa...just way too much fun.

on the bright side one can install gdb on the device
http://maemo.org/development/tools/#apt-example

set the repository up and
ssh into the device and
sudo apt-get install gdb

(hurah...ran gdb fennec (no debugging symbols) :( will look into this some more.)
and the file nsIdleServiceOSSO.cpp is not there to put a breakpoint on.
I think I'm missing something.

NSIdleService - test_bug343416.xul - rev2

2009-02-07T15:36:00.005-05:00

ref: http://zadkielm.blogspot.com/2009/02/nsidleservice-nsidleserviceossocpp.html

Along the process of finding out why nsIIdleService is not
working properly. Decided to start pokin' round in the .xul test.

Did a dump() on the variables...just to see what is going on.
http://mxr.mozilla.org/mozilla-central/source/widget/tests/test_bug343416.xul

nsIIdleService: nsIIdleService
nsIISCID: @mozilla.org/widget/idleservice;1
idleService: [xpconnect wrapped nsIIdleService @ 0x45aabb00 (native @ 0x45ab0d00)]
Try idleTime: 0
QueryInterface iid: {00000000-0000-0000-c000-000000000046}
nsISupports: true
nsIObserver: false
idleService.addIdleObserver: undefined
QueryInterface iid: {00000000-0000-0000-c000-000000000046}
nsISupports: true
nsIObserver: false
idleService.IdleObserver second time: undefined
QueryInterface iid: {00000000-0000-0000-c000-000000000046}
nsISupports: true
nsIObserver: false
idleService.removeIdleObserver: undefined
--DOMWINDOW == 12 (0x40320450) [serial = 14] [outer = 0x40320ae0] [url = about:blank]
newIdleTime: 0
var timeDiff: 5003
QueryInterface iid: {00000000-0000-0000-c000-000000000046}
nsISupports: true
nsIObserver: false

test_bug343416.xul with dump()
http://mozilla.pastebin.com/f2cd6d417

note: not enjoying that compiling everything is taking a while want to look in the nsIdleServiceOSSO.cpp

NSIdleService - nsIdleServiceOSSO.cpp - test_bug343416.xul

2009-02-06T09:02:00.004-05:00

There's plenty of them but this one uses the device
http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceOSSO.cpp

https://bugzilla.mozilla.org/show_bug.cgi?id=472548

note: learned xul dump();

looked into idleTime (which was failing)

--DOMWINDOW == 12 (0x40320b10) [serial = 13] [outer = 0x40320420] [url = about:blank]
++DOMWINDOW == 13 (0x40320b10) [serial = 15] [outer = 0x40320420]

idleTime: 0

--DOMWINDOW == 12 (0x403202a0) [serial = 14] [outer = 0x40320420] [url = about:blank]
--WEBSHELL 0x4033b660 == 5
--WEBSHELL 0x4033b850 == 4

ref: http://mxr.mozilla.org/mozilla-central/source/widget/src/gtk2/nsIdleServiceOSSO.cpp#66

returns a 0...so mIdle is not givin'r...which is connected to Observe.

http://mxr.mozilla.org/mozilla-central/source/widget/tests/test_bug343416.xul#46
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=477250

Who calls the Observe?

N810 setting up for Fennec rev3

2009-02-05T23:33:00.003-05:00

so for some reason...all the permissions issues just disappeared.
(maybe just needed to be turned off for a while...)

Ran the tests on the device.
Everything is going somewhat okay.
Got the expected results.

Just wished to be able to work from school...

Can run fennec via
Applications > Utilities > X Terminal

pretty spiffy...a bit slow but then again it's a debug build.

FOSSLC

2009-02-05T00:17:00.005-05:00

ref: http://www.fosslc.org/drupal/

After spending the a better part of the
day(actually 2 days now) trying to
figuring out what sort of issues
I needed to deal with setting up the N810
to be able to develop with Fennec.

I got an offer to go to a
mobile development conference
at Waterloo held by the

Free Open Source Software Learning Centre.
http://www.fosslc.org/drupal/node/172

It was interesting to see how things were similar and
different in the way one develops things on each platform.
They coverend iPhone, Android, and Open Moko.
It was a good balance and the conference could have
been much longer.

Found the Open Moko very interesting.
Since it felt very similar to how things are done
with Fennec and Maemo...minus the tools.

It was a good break from everything...

N810 setting up for Fennec rev2

2009-02-04T09:42:00.004-05:00

homeIP
Got it from
http://maemo.org/downloads/product/OS2008/homeip/
installed it and the applet doesn't show up on my Home screen...

note: needed to install python2.5-hildondesktop from
http://www.gronmayer.com/it/index.php?lang=en&system=maemo4&sort=hits&show_pck=126
gonna try to find another app that does the same thing...

ip address is in settings>Connection Manager>menu>Internet connection>IP address

Note: then you realise that Senenet is on a different network from the hardwired machines...sigh...will figure something out

N810 setting up for Fennec

2009-02-03T19:42:00.007-05:00

ref: https://wiki.mozilla.org/ReferencePlatforms/Test/Maemo

What is currently installed

open ssh
http://maemo.org/downloads/product/openssh/
homeip
not there

moredimmingoptions

nginx
ssh in as root and edit /etc/nginx/nginx.conf to change the default_type to text/plain.

fix apt-get

echo "deb http://p.quinput.eu/debfarm chinook user" >> /etc/apt/sources.list
apt-get update
apt-get -f install

install

apt-get install rsync wget bzip2 gnutar x11vnc python2.5-dev

x11vnc...not there


note: the ssh keeps cutting out...not sure if it's the local network
skipped buildbot and talos from the ref...transfers
are still going slowly over the network.

Getting a permission denied when running fennec stuff

not necessary to flash the device...unless your os is pre OS2008

Idle Time

2009-02-02T22:15:00.003-05:00

Fails on idle time

http://mozilla.pastebin.com/fd7ec2f

note:
http://mxr.mozilla.org/mozilla-central/source/widget/tests/test_bug343416.xul#100
https://developer.mozilla.org/en/nsIIdleService
https://developer.mozilla.org/en/nsIObserver

Configuring a device for Fennec

2009-02-02T09:59:00.002-05:00

ref: https://wiki.mozilla.org/ReferencePlatforms/Test/Maemo

So there are instruction there...
gonna giv'r with them hopefully I can get something working from it.

note: GDB's reliability on the device is currently unknown. However
it is possible to run individual chrome tests via command line(do not
know how that is done yet). Apparently there will need to be some
funky stuff to get the http sever going and all the paths correct.

side note: this is turning out to be an interesting adventure

chrome_content.log - a first look

2009-02-01T15:00:00.005-05:00

file content_chrome.log :
http://mozilla.pastebin.com/fd7ec2f

console output:
http://mozilla.pastebin.com/f2d1a569d

Failure Notes:
chrome://mochikit/content/chrome/widget/test/test_bug343416.xul:

nsIIdleService should exist and be implemented on all tier 1 platforms.
Getting the idle time should not fail in normal circumstances on any tier 1 platform.
The idle time should have increased by roughly the amount of time it took for the timeout to fire.
We added a listener and it should have been called by now.

ref: https://bugzilla.mozilla.org/show_bug.cgi?id=472548#c3

Getting Tests to run rev2 - goin in the right direction

2009-01-31T22:48:00.004-05:00

Once the 3 patches were installed and
470971, 460515, and 421611

followed the instructions
ref:
https://bugzilla.mozilla.org/show_bug.cgi?id=421611#c40

ported the tests over

https://wiki.mozilla.org/Mobile/Fennec_Mochitest#Porting_tests_to_N800
note: make make package-test not make package

ran CHROME test

python mochitest/runtests.py --appname=/media/mmc1/zad/fennec/fennec --chrome --utility-path=/media/mmc1/zad/bin --certificate-path=/media/mmc1/zad/bin --xre-path=/media/mmc1/zad/fennec/xulrunner --log-file=chrome_content.log --autorun --close-when-done

stuff happened...apparently correct stuff(lately it's been the incorrect stuff)
things are somewhat goin' spiffy...

note: there's a log file that I forgot to check will do it soon.

Getting Tests to run

2009-01-30T20:24:00.010-05:00

once you have .hgrc setup with hg queues

hg qinit -c

in your mozilla-central folder

hg pull;hg update;cd mobile;hg pull;hg udpate; cd ..
hg qapplied --- should not say anything
hg qnew gredir
patch -p1 < Bug470971.txt
hg qrefresh
hg qnew xpcshell
patch -p1 < Bug460515.txt
hg qrefresh
hg qnew mochitest
patch -p1 < Bug421611.txt
hg qrefresh
hg qapplied <- should be the three patches

note: permissions issues keeps coming up and wont build...maybe fakeroot might fix it

updated steps to get tests to work on a device.
https://bugzilla.mozilla.org/show_bug.cgi?id=421611#c40
setup which chrome tests (used widget instead of content)
http://zadkielm.blogspot.com/2009/01/fennec-and-chrome-testing.html
plus to get tests over
https://wiki.mozilla.org/Mobile/Fennec_Mochitest#Porting_tests_to_N800

got the tests over to device and untar'ed it...
/media/mmc1/zad/

Nokia-N810-51-3:/media/mmc1/zad# ls
bin certs fennec mochitest plugins tarfiles

note: how do you tell if fennec ran properly in from an ssh connection

will run this command once files are copied over.

Nokia-N810-51-3:/media/mmc1/zad/mochitest# python runtests.py --appname=../fennec/fennec --chrome --log-file=chrome_content.log --autorun --close-when-done

error
Traceback (most recent call last):
File "runtests.py", line 549, in
main()
File "runtests.py", line 291, in main
server.start()
File "runtests.py", line 202, in start
self._process = automation.Process([xpcshell] + args, env = env)
File "/usr/lib/python2.5/subprocess.py", line 594, in __init__
File "/usr/lib/python2.5/subprocess.py", line 1091, in _execute_child
OSError: [Errno 2] No such file or directory
Nokia-N810-51-3:/media/mmc1/zad/mochitest#

Need to pass utility-path, certificate-path, and xre-path to the command line

python mochitest/runtests.py --appname=/root/fennec/fennec --utility-path=/root/bin --certificate-path=/root/bin --xre-path=/root/fennec/xulrunner --test-path=MochiKit_Unit_Tests --autorun

hg queues in scratchbox

2009-01-29T23:03:00.007-05:00

/scratchbox/login

.hgrc
must be placed in
~/

~/.hgrc

[ui]
username = Username
merge = diff

[extensions]
hgext.mq =

[diff]
git = 1
showfunc=True

[defaults]
diff=-p -U 8

to initialise queues
in your source folders. (mozilla-central)

hg qinit -c

note: there's 2 sources for fennec mozilla-central and mobile(which is inside mozilla-central) it might get confusing applying patches on the two...hg qapplied will be useful

note: trying to fix my other install of scratchbox on another VM...hopefully i have uninstalled and reinstalled properly.

So why can't we run tests?

2009-01-28T20:41:00.003-05:00

This patch is the reason why things are some what more challenging than usual.
https://bugzilla.mozilla.org/show_bug.cgi?id=470963

3 patches are needed in order to get tests to work on a device
note: we will to setup hg queues in scratchbox setup an .hgrc
in ~/.

Need patches from bug 470971, 460515, and 421611

Has some instructions on how to get it goin' again...will giv'r with soon.
https://bugzilla.mozilla.org/show_bug.cgi?id=421611#c29

Thing will probably change over the next 24 hrs....so will need to get
an update before we'll giv'r.

Fennec build with absolute paths...[0.4]

2009-01-27T14:55:00.014-05:00

So not sure why my build is giving me absolute paths
inside automation.py...I think I did things correct.

setting up scratchbox
http://zadkielm.blogspot.com/2009/01/scratchbox-on-32-bit-ubuntu.html
and building fennec
http://zadkielm.blogspot.com/2009/01/building-fennec.html

However
from mobile_debug

xulrunner/_tests/testing/mochitests/automation.py
http://mozilla.pastebin.com/f5bb7da5c

I am getting absolute paths and missing files.

DIST_BIN = "/home/seneca/mozilla/mobile_debug/xulrunner/dist/bin" (this one can b changed)
.
.
.
CERTS_SRC_DIR = "/home/seneca/mozilla/mozilla-central/build/pgo/certs" (i dont know where this is)

note: unable to run chrome tests on a device if paths are absolute

tried here...no luck
[sbox-CHINOOK_ARMEL: ~/mozilla/mobile_debug/xulrunner/_profile/pgo] > ls
automation.py favicon.ico index.html profileserver.py
blueprint genpgocert.py js-input server-locations.txt

untarred fennec-linux-arm (missing pgo certs...so it's not there either.)
http://mozilla.pastebin.com/f311b1256

A little confused as to what is going on...did this twice on 2 different Ubuntu VMs

filed a bug: Bug 475586
ref - Allow XULRunner apps to use Mochitest framework in 1.9.1
https://bugzilla.mozilla.org/show_bug.cgi?id=460515
ref - Remove assumption that xpcshell etc in same directory as app executable
https://bugzilla.mozilla.org/show_bug.cgi?id=460515

one step forward...

2009-01-26T22:43:00.009-05:00

For some reason errors plague my build.

click for more information
http://groups.google.com/group/mozilla.dev.builds/browse_thread/thread/ea3d104fd39776ac#

note: so close this morning...had a linux arm build copied over to the device...then found out that that the build didn't have the proper test folders...and it was using absolute paths in automation.py...then run into these permission errors...thought it was my scratchbox build...so reinstalled it again...then the school's internet kept cutting out...then still the same errors...did find out where the libIDL-config was and exported it to PKG_CONFIG_PATH.

note: might have sbox_adduser as user and not as root. Couldn't find a way to fix it or reverse it. installed a new vm and giv'rd again.

note: 6:30 am the next day got it build still using absolute paths in ../mochitests/automation.py ...

for future ref:
https://wiki.mozilla.org/Mobile/Notes
https://wiki.mozilla.org/Mobile/DFBPorting
http://groups.google.com/group/mozilla.dev.platforms.mobile/topics

Uninstall Scratchbox

2009-01-25T18:56:00.006-05:00

ref: http://www.scratchbox.org/documentation/user/scratchbox-1.0/html/installdoc.html#AEN610
note: got way to many errors my first try so uninstalled and trying again# apt-get remove scratchbox-libs
removes all info int he users folder(save all your important files)

# /scratchbox/sbin/sbox_ctl stop

/scratchbox directory should be empty# rm -rf /scratchbox/users
scratchbox mounts directories sbox_ctl stop then you can remove the rest of scratchbox
# rm -r /scratchbox

Porting Tests to a Device

2009-01-24T21:00:00.006-05:00

ref: https://wiki.mozilla.org/Mobile/Fennec_Mochitest#Porting_tests_to_N800

local machine in path/to/mobile_debug (mobilebase)

cd xulrunner
tar chf tests.tar _tests/ _profile/

inside your fennec folder of the device untar tests

mv xulrunner bin
mkdir xulrunner; mkdir xulrunner/dist
mv bin xulrunner/dist/bin
mv _tests xulrunner/_tests
mv _profile xulrunner_profile

then follow chrome testing instructions
http://zadkielm.blogspot.com/2009/01/fennec-and-chrome-testing.html

note: for Bug 472548 you want to use

cp -R chrome.bak/widget chrome/
instead of content

Bugs (nsIdleService, pause) and Osos (OSSO)

2009-01-23T11:58:00.010-05:00

A summary so far...

ref: https://bugzilla.mozilla.org/show_bug.cgi?id=448544
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=472548

There is a nsIdleServiceOSSO.cpp that
might be helpful to tie the two bugs together.
(has a vague idea)

note: setting things up takes time...and there is a lot
of information to take in(click: idle pause)

OSSO what does it stand for?
Open Source Software Open?

OSSO used to be the open source unit in nokia,

Bug 472548 is failing maemo tests using nsIdleService.

/mozilla/widget/src/xpwidgets/nsIdleService.cpp

when it is testing on maemo.

/mozilla/widget/src/gtk2/nsIdleServiceOSSO.cpp

thinking...
OSSO > nokia > maemo

What is nsIdleServiceOSSO.cpp?

from moznet #mobile
<tonikitoo> right, so system broadcast some messages , like system-idle and so on. this file listen to those and take the proper action.
like it could stop all plugins, or wahtever

nsIdleServiceOSSO.cpp

Images that wont pause.
Bug 448544
https://bugzilla.mozilla.org/show_bug.cgi?id=448544

there's a patch...but it stops and restarts not pauses.

Updating Fennec Source

2009-01-22T22:30:00.003-05:00

It's as easy as up, up, down, down, left, right, left, right, select, start!

There are 2 repositories...mozilla-central and mobile

starting from "mozilla-central" repository folder(hg clone http://hg.mozilla.org/mozilla-central)
hg qpop -a;hg pull;hg update; cd mobile;hg pull;hg update; cd ..;hg qpush -a

hg qpop -a (removes all the current patches you have in queue...see hg queues)
hg pull (update the source)
hg update
cd mobile (switch to the mobile repository)
hg pull (update the source)
hg update
cd ..
hg qpush -a (push all of your patches)

Colour reftesting Bug 463221

2009-01-21T22:48:00.003-05:00

ref: https://bugzilla.mozilla.org/show_bug.cgi?id=463221

The tests checks to see if the colours with ICC profiles
(Adobe1998, ColorMatch, ProPhoto)
renders properly in the browser.

It failed on Linux due to a preprocessor issue
with the lcms not being able to detect for an x86
configuration.

When this was fixed Re-tested on Mac, Linux, Windows and all passed.

Hurah.

The bits of Fennec for a device

2009-01-21T22:05:00.014-05:00

So in the scratchbox environment.

/scratchbox/login

To build the bits necessary for Fennec that will go on your device (like an Nokia N810)

cd path/to/mobilebase/mobile

then
run

make package

you should have a file in your

mobilebase/mobile/dist

called

fennec-1.0a3*.tar.bz2
in in case
fennec-1.0a3pre.en-US.linux-arm.tar.bz2
(if it's not linux-arm then your didn't set your environment right...uninstall and try again)

copy to device using secure copy

scp fennec-1.0a3pre.en-US.linux-arm.tar.bz2 root@ip.to.the.device:/path/to/folder

bunzip2 filename

note: nokia 810 can't untar .bz2...untar on your machine before copying to the device.

scp -r fennec root@ip.to.the.device:/path/to/folder

-r is for folders

note: this might be slow depending on your connection

scratchbox, sdk - cs2007q3 toolchain on 32-bit Ubuntu

2009-01-20T22:07:00.023-05:00

ref:
https://wiki.mozilla.org/ReferencePlatforms/Linux-CentOS-5.0#Installing_Scratchbox

ref: http://www.scratchbox.org/ (information about scratchbox)

In order to setup scratchbox you will need to install
scratchbox, scratchbox-toolchain, scratchbox-sdk, and libIDL.
These instructions will guide you through all the steps
in order to setup your scratchbox environment
to build Fennec...

1
add the following lines to /etc/sysctl.conf

vm.vdso_enabled = 0
vm.mmap_min_addr = 4096

Then run

sudo sysctl -p

2
As ROOT:
cd /
wget http://repository.maemo.org/stable/chinook/maemo-scratchbox-install_4.0.1.sh
chmod a+x ./maemo-scratchbox-install_4.0.1.sh
./maemo-scratchbox-install_4.0.1.sh -s /scratchbox
(depending on your connection this download will take a while)
/scratchbox/sbin/sbox_adduser USERNAME(ex. seneca) yes
note: very important to do this as root I don't know how to reverse
it if you do it as a user.

Note: you might need to change permissions in the /scratchbox/

3
ref: https://wiki.mozilla.org/index.php?title=Mobile/Build/Maemo_Build_Instructions&oldid=117369#Setting_Up_the_Maemo_Chinook_SDK

install cs2007q3 toolchain
sudo su
echo "deb http://scratchbox.org/debian/ stable main" >> /etc/apt/sources.list
apt-get update
apt-get install scratchbox-toolchain-cs2007q3-glibc2.5-arm6
install maemo sdk
wget http://tablets-dev.nokia.com/4.0.1/maemo-sdk-install_4.0.1.sh
chmod +x maemo-sdk-install_4.0.1.sh
exit root
bash maemo-sdk-install_4.0.1.sh
note: theres an installer(choose to install everything)

install libIDL (fennec needs this to build)
in scratchbox (/scratchbox/login) run
wget http://ftp.gnome.org/pub/gnome/sources/libIDL/0.8/libIDL-0.8.9.tar.gz
tar -xzvf libIDL-0.8.9.tar.gz
cd libIDL-0.8.9
./configure
make
fakeroot make install
cd ..

note: if the libIDL error appears on build run this.
export PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pgkconfig

4
Some Scratchbox Commands:

ADD a user(done as root(#)):

# /scratchbox/sbin/sbox_adduser USERNAME yes

Scratchbox service must be started for CPU transparency to be functional.
Run the following command (outside scratchbox with root permissions):

# /scratchbox/sbin/sbox_ctl start (if you get an ERROR: Scratchbox is not properly set up! run this)


Now you are ready to build Fennec on scratchbox...it's the same instructions.

Building Fennec Mac

2009-01-18T19:08:00.006-05:00

ref: http://zadkielm.blogspot.com/2009/01/building-fennec.html

get the source

hg clone http://hg.mozilla.org/mozilla-central
cd mozilla-central
hg clone http://hg.mozilla.org/mobile-browser mobile

.mozconfig used

# Options for client.mk.
mk_add_options MOZ_BUILD_PROJECTS="xulrunner mobile"
mk_add_options MOZ_OBJDIR=@TOPSRCDIR@/../mobilebase

# Global options
ac_add_options --enable-debug
ac_add_options --disable-optimize
ac_add_options --disable-crashreporter

# XULRunner options
ac_add_app_options xulrunner --enable-application=xulrunner
ac_add_app_options xulrunner --disable-javaxpcom

# Enabling --with-arm-kuser implies Linux on ARM and enables kernel
# optimizations for that platform
#ac_add_app_options xulrunner --with-arm-kuser

# Disabled tests due to bug 454881
ac_add_options --enable-tests
ac_add_options --enable-mochitest

# mobile options
ac_add_app_options mobile --enable-application=mobile
ac_add_app_options mobile --with-libxul-sdk=../xulrunner/dist

build

make -f client.mk build

Does it build? Yes.

Where to find it?
mobilebase/mobile/dist/FennecDebug.app/Contents/MacOS

Can it run chrome/mochi tests?
Running into: No such file or directory
error output:
http://mozilla.pastebin.com/f6e047dbe

Fennec and Chrome testing

2009-01-17T21:31:00.003-05:00

What are Chrome tests?
ref: https://developer.mozilla.org/En/Chrome_tests

How do one runs Chrome tests on Fennec.
ref: https://wiki.mozilla.org/Mobile/Fennec_Chrome

from path to/mobilebase/xulrunner/

cd _tests/testing/mochitest

## create backup
cp -R chrome chrome.bak

## NOTE: you need to do the following steps in a look for each directory
## dirs: [content, docshell, dom, toolkit, widget]
rm -Rf chrome/*

cp -R chrome.bak/content chrome/

python runtests.py --appname=path/to/fennec --chrome --log-file=chrome_content.log --autorun --close-when-done

Fennec and Mochihitests

2009-01-17T21:16:00.003-05:00

What are Mochitests?
ref: https://developer.mozilla.org/en/Mochitest

Running Mochitests on Fennec.
in your .mozconfig

add these lines

ac_add_options --enable-tests
ac_add_options --enable-mochitest

build it again
and to run the tests (from pathtosource/mobilebase)

cd xulrunner/_tests/testing/mochitest
python runtests.py --appname=
ex. python runtests.py --appname=/home/seneca/mozilla/mobilebase/mobile/dist/bin/fennec
note: no space between --appname=/home

Building Fennec

2009-01-17T20:41:00.008-05:00

notes: key is repetition...

What is Fennec?
ref: http://en.wikipedia.org/wiki/Fennec_(browser)

How do you Build it?
ref: https://wiki.mozilla.org/Mobile/Build/Fennec

Steps are done on Ubuntu.

pull the source using hg (Mercurial)

hg clone http://hg.mozilla.org/mozilla-central
cd mozilla-central
hg clone http://hg.mozilla.org/mobile-browser mobile

cd mozilla-central
setup the .mozconfig file

sample:

# Options for client.mk.
mk_add_options MOZ_BUILD_PROJECTS="xulrunner mobile"
mk_add_options MOZ_OBJDIR=@TOPSRCDIR@/../mobilebase

# clint additions
ac_add_options --enable-debug
ac_add_options --disable-optimize
ac_add_options --disable-crashreporter
#ac_add_options --disable-dbus

ac_add_options --enable-tests
ac_add_options --enable-mochitest

# XULRunner options
ac_add_app_options xulrunner --enable-application=xulrunner
ac_add_app_options xulrunner --disable-javaxpcom

# mobile options
ac_add_app_options mobile --enable-application=mobile
ac_add_app_options mobile --with-libxul-sdk=../xulrunner/dist

How to build Fennec?

make -f client.mk build

Where did it go?

cd ../mobilebase/mobile/

How to run(from mobile)

dist/bin/fennec

Hi Fennec.

it's pretty spiffy...

Note: while compiling...it might look like it stalled...it didn't

When Bugs come in Pairs...

2009-01-17T18:00:00.005-05:00

After several days of searching for a project for DPS911.
It has been narrowed down to these two Bugs.

Bug 472548 - nsIdleService tests fail on maemo
Bug 448544 - Pause image animation when system idle.

Everyone was very helpful during the search.
Spent several days parsing through
information and what not.

These two bugs are Fennec related
Currently the browser doesn't pause
image animation when the system is idle.
With the idle service failing(on maemo)
there is no way to detect if the system
is idle to pause the image.

So one has to fix one in order
to get the other to work.

Contact:
Joel Maher
Time to setup the Fennec...

Tracking down bug 459617 rev4

2009-01-12T14:41:00.003-05:00

So after applying the patch
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=456028
ref: https://bugzilla.mozilla.org/attachment.cgi?id=356334

the failing pixel passes on an __x86_64__
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=459617

Tracking down bug 459617 rev3

2009-01-09T10:37:00.004-05:00

ref:
http://zadkielm.blogspot.com/2009/01/tracking-down-bug-459617-rev2.html
http://mozilla.pastebin.com/f5093919d

made changes in the preprocessor in

http://mxr.mozilla.org/mozilla-central/source/modules/lcms/src/cmsxform.c#96

note: Sleeping for 300 seconds means crashes

Crashes:

Program received signal SIGSEGV, Segmentation fault.
0x00007f763c9b67c4 in MatrixShaperXFORMFloat (p=0x7f7643450fe0, in=0x7f7637bd6c21, out=0x7f7637bd6c21, Size=10)
at /home/seneca/ffsrc/modules/lcms/src/cmsxform.c:764
764 Out[0] = MatShaper->L_Precache->Impl.LI168_REVERSE.Cache[0][tmp[0]];
(gdb) bt
#0 0x00007f763c9b67c4 in MatrixShaperXFORMFloat (p=0x7f7643450fe0, in=0x7f7637bd6c21, out=0x7f7637bd6c21, Size=10)
at /home/seneca/ffsrc/modules/lcms/src/cmsxform.c:764
#1 0x00007f763c9b8fee in cmsDoTransform (Transform=0x7f7643450fe0, InputBuffer=0x7f7637bd6c21, OutputBuffer=0x7f7637bd6c21, Size=10)
at /home/seneca/ffsrc/modules/lcms/src/cmsxform.c:2002
#2 0x00007f7638eee480 in row_callback (png_ptr=0x7f763af56000,
new_row=0x7f7637bd6c21 "\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233", '�' , "\020ټ8v\177", row_num=0, pass=0) at /home/seneca/ffsrc/modules/libpr0n/decoders/png/nsPNGDecoder.cpp:761
#3 0x00007f7638f1ff4c in MOZ_PNG_push_have_row (png_ptr=0x7f763af56000,
row=0x7f7637bd6c21 "\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233\214�\233", '�' , "\020ټ8v\177")
at /home/seneca/ffsrc/modules/libimg/png/pngpread.c:1742
#4 0x00007f7638f1fad0 in MOZ_PNG_push_proc_row (png_ptr=0x7f763af56000) at /home/seneca/ffsrc/modules/libimg/png/pngpread.c:1167
#5 0x00007f7638f1f3e8 in MOZ_PNG_proc_IDAT_data (png_ptr=0x7f763af56000, buffer=0x7f763ade71a8 "\030\031c�Y<\233\0017`�-\005\222\031��", buffer_length=22) at /home/seneca/ffsrc/modules/libimg/png/pngpread.c:984

this bug is currently being blocked by
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=456028
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=459617#c16
ref: https://bugzilla.mozilla.org/show_bug.cgi?id=456028#c7

finding a way to compile a 32-bit ver of firefox on the ubuntu
Tested on i686 and it passes

ref: https://bugzilla.mozilla.org/show_bug.cgi?id=459617#c8
ref: https://developer.mozilla.org/en/Compiling_32-bit_Firefox_on_a_Linux_64-bit_OS
no success

going to try smartpm, alien to install rpms

was suggested to run a VM however it creates issues with the colour reftest.
ref: http://zadkielm.blogspot.com/2008/11/colour-profile-reftests-on-virtual.html

Tracking down bug 459617 rev2

2009-01-05T06:59:00.003-05:00

ref: http://zadkielm.blogspot.com/2009/01/tracking-down-bug-459617-rev1.html

HAVE_SSE2_INTEL_MNEMONICS is not defined (determines which floating point path to use)
ref: http://mxr.mozilla.org/mozilla-central/source/modules/lcms/src/cmsxform.c#1566

GDB MAC

gdb MAC on 'b cmsxform.c:1563'

Breakpoint 1, PickTransformRoutine (p=0x1ba13120, FromTagPtr=0xbfffce40, ToTagPtr=0xbfffce3c) at /Users/evmarbella/mozilla/ffsrc/modules/lcms/src/cmsxform.c:1568
1568 if (!SSE2Available())

GDB LINUX

gdb LINUX on 'b cmsxform.c:1563'

Breakpoint 1, PickTransformRoutine (p=0x7f947a850fe0, FromTagPtr=0x7fff8b55b1cc, ToTagPtr=0x7fff8b55b1c8) at /home/seneca/ffsrc/modules/lcms/src/cmsxform.c:1565
1565 p -> dwOriginalFlags &= ~cmsFLAGS_FLOATSHAPER;

Leads to a Preprocessor issue where
ref: http://mxr.mozilla.org/mozilla-central/source/modules/lcms/src/cmsxform.c#64

did an
#ifdef __GNUC__
printf(..._
#ifdef __I386__
printf(..._

Linux only
defined __GNUC__
not defined__i386__

Mac defines them both.

The wrong floating point(HAVE_SSE2_INTEL_MNEMONICS),
wrong transform (p->xform) leads to the colour reftests are failing in Linux.

see comment: https://bugzilla.mozilla.org/show_bug.cgi?id=459617#c4

Tracking down bug 459617 rev1

2009-01-03T10:21:00.007-05:00

ref.
http://zadkielm.blogspot.com/2008/12/tracking-down-bug-459617.html

Checked to see if
gCMSsRGBProfile in

http://mxr.mozilla.org/mozilla-central/source/gfx/thebes/src/gfxPlatform.cpp#641
and the colour profile in the reftest
were the same.

output Ubuntu
gCMSsRGBProfile: 81559000

The colour we care about rgb 140,163,155 - 8ca39b was Transformed to
r - 131
g - 164
b - 156

p->OutputProfile 81559000

output Mac

gCMSsRGBProfile: AF3600

The pixel we care about (rgb 140,163,155 - 8ca39b) was Transformed to
r - 130
g - 164
b - 156

output profiles compared 'p' sRGB AF3600

So it's the same
then checked to see what xform it's using in the structure

p->xform(...)
http://mxr.mozilla.org/mozilla-central/source/modules/lcms/src/cmsxform.c#1971

~/mozilla/ffsrc/obj-ff/dist/MinefieldDebug.app/Contents/MacOS/firefox -P TestFirefox003 -g -reftest ./failedPixeltestAdobe1998.list
note: -g runs things in gdb
b cmsxform.c:1997
run
disp p

MAC debug:
2: *p = {
InputFormat = 262169,
OutputFormat = 262169,
StrideIn = 0,
StrideOut = 0,
Intent = 0,
ProofIntent = 3,
DoGamutCheck = 0,
InputProfile = 0xafc000,
OutputProfile = 0xaf3400,
PreviewProfile = 0x0,
EntryColorSpace = icSigRgbData,
ExitColorSpace = icSigRgbData,
dwOriginalFlags = 384,
m1 = {
v = {{
n = {65536, 0, 0}
}, {
n = {0, 65536, 0}
}, {
n = {0, 0, 65536}
}}
},
m2 = {
v = {{
n = {0, 0, 0}
}, {
n = {0, 0, 0}
}, {
n = {0, 0, 0}
}}
},
of1 = {
n = {0, 0, 0}
},
of2 = {
n = {0, 0, 0}
},
xform = 0x11b72021 <MatrixShaperXFORMFloat>,
FromInput = 0x11b6b400 <Unroll3Bytes>,
FromDevice = 0,
Stage1 = 0,
Stage2 = 0,
ToDevice = 0,
ToOutput = 0x11b6cb24 <Pack3Bytes>,
Device2PCS = 0x0,
PCS2Device = 0x0,
Gamut = 0x0,
Preview = 0x0,
DeviceLink = 0x0,
GamutCheck = 0x0,
InMatShaper = 0x0,
OutMatShaper = 0x0,
SmeltMatShaper = 0x1a6069d0,
Phase1 = 0,
Phase2 = -1,
Phase3 = 0,
NamedColorList = 0x0,
lInputV4Lab = 0,
lOutputV4Lab = 0,
CacheIn = {0 <repeats 16 times>},
CacheOut = {0 <repeats 16 times>},
AdaptationState = 0,
rwlock = 0
}

Linux Output:

1: *p = {
InputFormat = 262169,
OutputFormat = 262169,
StrideIn = 0,
StrideOut = 0,
Intent = 0,
ProofIntent = 3,
DoGamutCheck = 0,
InputProfile = 0x7fc91826d000,
OutputProfile = 0x7fc91ab59000,
PreviewProfile = 0x0,
EntryColorSpace = icSigRgbData,
ExitColorSpace = icSigRgbData,
dwOriginalFlags = 256,
m1 = {
v = {{
n = {65536, 0, 0}
}, {
n = {0, 65536, 0}
}, {
n = {0, 0, 65536}
}}
},
m2 = {
v = {{
n = {0, 0, 0}
}, {
n = {0, 0, 0}
}, {
n = {0, 0, 0}
}}
},
of1 = {
n = {0, 0, 0}
},
of2 = {
n = {0, 0, 0}
},
xform = 0x7fc91cdb71c1 <MatrixShaperXFORM>,
FromInput = 0x7fc91cdb0b0c <,
FromDevice = 0,
Stage1 = 0,
Stage2 = 0,
ToDevice = 0,
ToOutput = 0x7fc91cdb212e <Pack3Bytes>,
Device2PCS = 0x0,
PCS2Device = 0x0,
Gamut = 0x0,
Preview = 0x0,
DeviceLink = 0x0,
GamutCheck = 0x0,
InMatShaper = 0x0,
OutMatShaper = 0x0,
SmeltMatShaper = 0x7fc91bf721a0,
Phase1 = 0,
Phase2 = -1,
Phase3 = 0,
NamedColorList = 0x0,
lInputV4Lab = 0,
lOutputV4Lab = 0,
CacheIn = {0 <repeats 16 times>},
CacheOut = {0 <repeats 16 times>},
AdaptationState = 0,
rwlock = 0
}
MAC uses xform = 0x11b72021 matrixshaperxformfloat ,
LINUX uses xform = 0x7fc91cdb71c1 matrixshaperxform ,

should be float for both.

Row1 All RGB Data...failing colours

2008-12-28T11:21:00.006-05:00

image ref: http://www.hotpng.com/gallery/main.php

started looking at how to find how many colours are failing

sampled an image with all 16,777,216 RGB colours
applied adobe1998 to all images via Automator script
Then ran failed colour reftests to get firefox rgb outputs
then ran a test that compares images with profiles against
the output.

tested row1 of 16

col1 - 16

(left - right)

findings:

ROW1 differing pixels TOTAL:

col16 98
col15 531
col14 512
col13 512
col12 50
col11 1225
col10 2429
col09 835
col08 743
col07 1434
col06 149
col05 795
col04 8
col03 122
col02 422
col01 520

TOTAL 10385 of 1,048,576 unique colours on row 1 with differing pixels

#row1 - 256X256 images w/ 65,536 colours each x 16 = 1,048,576 unique colours per row.

note: tried to run the test and grepp'ing
for differing pixels however it randomly misses some
why is that? gotta do it one at a time to to make sure it's right.

see image below highlighted row1...columns are squares broken up 256x256

Results:
http://pastebin.mozilla.org/600543
http://matrix.senecac.on.ca/~evmarbella/Row1.zip

Finding the failing range of pixels in Colour Reftest

2008-12-24T10:00:00.011-05:00

refences:

http://zadkielm.blogspot.com/2008/12/re-thinking-color-reftests.html

Above Ubuntu and Mac difference/levels adjustment note easy to tell which colours failed.

Above Difference/levels adjustment between Mac output and Fedora output.
Note: there's a difference but it seems to be all of it.
REFTEST number of differing pixels: 356 out of 65,536(unique colors) < how many pixels failed?, which ones? can we compare?

Above Firefox colour reftest Fedora8 output

Above is a Firefox colour reftest Mac output.

re-thinking color reftests

2008-12-23T11:27:00.009-05:00

references colour reftests and difference tests from:
http://zadkielm.blogspot.com/2008/12/finding-failing-range-of-pixels-in.html
http://zadkielm.blogspot.com/2008/12/tracking-down-bug-459617.html

Currently there is an issue where the lcms inside firefox is outputting an off-by-one on colour reftests in the red channel. Currently it is tricky to detect which pixels are failing since it is a range of them. The initial plan was to take +16Million colours and do a colour reftest on all of them. Find out the range and fix it. However finding out which pixels are different is an issue. Tried to do a difference on the two images however it is not narrowing down which pixels they are.

Question: Is there a way to get the range of pixels in the reftest?
note: the colour reftest indicates how many pixels failed.

output ON FEDORA:

REFTEST number of differing pixels: 356 < how many pixels failed?, which ones? can we compare?
REFTEST FINISHED: Slowest test took 2807ms (file:///home/ezadkiel/color-profile/pngfirefoxrenders/Row1-Col1-AdobeRGB1998-ffoutput.png)
REFTEST INFO | Result summary:
REFTEST INFO | Successful: 0 (0 pass, 0 load only)
REFTEST INFO | Unexpected: 1 (1 unexpected fail, 0 unexpected pass, 0 failed load, 0 exception)
REFTEST INFO | Known problems: 0 (0 known fail, 0 random, 0 skipped)
REFTEST INFO | Total canvas count = 2

Objective-C Date and Times to String

2008-12-15T15:04:00.004-05:00

ref:
http://developer.apple.com/documentation/Cocoa/Conceptual/DatesAndTimes/Articles/dtDatesToStrings.html

formatting Dates

Conversion Specifier	Description
`%%`	a '`%`' character
`%a`	abbreviated weekday name
`%A`	full weekday name
`%b`	abbreviated month name
`%B`	full month name
`%c`	shorthand for `%X %x`, the locale format for date and time
`%d`	day of the month as a decimal number (01-31)
`%e`	same as `%d` but does not print the leading 0 for days 1 through 9
`%F`	milliseconds as a decimal number (000-999)
`%H`	hour based on a 24-hour clock as a decimal number (00-23)
`%I`	hour based on a 12-hour clock as a decimal number (01-12)
`%j`	day of the year as a decimal number (001-366)
`%m`	month as a decimal number (01-12)
`%M`	minute as a decimal number (00-59)
`%p`	AM/PM designation for the locale
`%S`	second as a decimal number (00-59)
`%w`	weekday as a decimal number (0-6), where Sunday is 0
`%x`	date using the date representation for the locale
`%X`	time using the time representation for the locale
`%y`	year without century (00-99)
`%Y`	year with century (such as 1990)
`%Z`	time zone name (such as Pacific Daylight Time)
`%z`	time zone offset in hours and minutes from GMT (HHMM)

Tracking down bug 459617

2008-12-15T08:39:00.009-05:00

references:
https://bugzilla.mozilla.org/show_bug.cgi?id=459617
http://zadkielm.blogspot.com/2008/12/reftest-03-manifesting-bug-459617.html

changed the
cmsxform.c
to test for the failed pixels

http://pastebin.mozilla.org/592294

Mac OSX output:
The pixel we care about (rgb 140,163,155 - 8ca39b) was Transformed to
r - 130
g - 164
b - 156

Ubuntu output:
The pixel we care about (rgb 140,163,155 - 8ca39b) was Transformed to
r - 131 <------off by one
g - 164
b - 156

Fedora output:
The pixel we care about (rgb 140,163,155 - 8ca39b) was Transformed to
r - 131 <------off by one
g - 164
b - 156

Note:
It is not just one pixel but a range of pixels...tracking it down might be tricky.

Xcode Disable Assertions on Release

2008-12-13T23:15:00.002-05:00

In a project there are 2 build configurations:
Debug
Release
in a Debug we usually want Assertions
in a Release we don't want Assertions

to block assertions on Release
double click on your main project file
(navigate to your project info)
under Build
Select
Configuration > Release
Scroll down to
GCC 4.0 Preprocessing
then under
Preprocessor Macros add:
NS_BLOCK_ASSERTIONS

Xcode Debug

2008-12-13T22:47:00.005-05:00

trying out Xcode and Objective-C
The first time debugging it didn't want to hit any breaks points...
did a
Build > Clean All
then
Build and Go
seems to work swell...
Xcode is so fancy...with all its Huds and what not.
and it uses gdb and gcc.
Access the Debug Console and all the gdb commands works there.

What's A Manifest? - reftesting and manifests

2008-12-08T15:25:00.004-05:00

The "manifest" refers to the reftest.list file.

reftest.list = manifest

It can mark a file as a known failure using the flags described here:
http://mxr.mozilla.org/mozilla-central/source/layout/tools/reftest/README.txt
the read me can be found here.

reftest 0.3 manifesting bug 459617

2008-12-06T14:30:00.007-05:00

Made changes to the tests
and put up an updated patch

https://bugzilla.mozilla.org/show_bug.cgi?id=463221

but due to 3.1 branching
and a bug that was found the tests cannot be landed
or place on the manifest fail-on-linux
note: what/where is a manifest?! dec 6. 08

So onwards to the bug and trying to track it down
steps:

1. figure out the problematic pixel

note: there are many of them
8ca39b with colour profile Adobe1998.png
6f8753 with colour profile Adobe1998.png

note: the image all the visible dots are failed pixels

image above is a blend mode Difference between
an image with a colour profile(adobe1998) and the
reftest rendered image(forced rgb).

2.add conditional code in cmsDoTransform to check for that pixel,
prints it on input
and prints the transformed pixel

note: started to track track this down on the mxr...this might be it

http://mxr.mozilla.org/mozilla-central/source/modules/lcms/src/cmsxform.c#1965

1964 // Apply transform code
1965 void LCMSEXPORT cmsDoTransform(cmsHTRANSFORM Transform,
1966 LPVOID InputBuffer,
1967 LPVOID OutputBuffer, unsigned int Size)
1968
1969 {
1970
1971 _LPcmsTRANSFORM p = (_LPcmsTRANSFORM) (LPSTR) Transform;
1972
1973 p -> StrideIn = p -> StrideOut = Size;
1974
1975 p -> xform(p, InputBuffer, OutputBuffer, Size);
1976
1977 }

Step2 would confirm if it is an LCMS issue or not
note: working on step2...it seems to call this on every image

If it is
get the lcms code to say that it's outputting different
pixels on different platforms for the same input
pixel and transform.

Build Issues Bug (builds like it's linux)

2008-12-06T13:01:00.003-05:00

From:
http://zadkielm.blogspot.com/2008/11/building-issues-on-my-mac-os-x.html

Had build issues coming from the .mozconfig on a mac build
it thought it was linux the error went by so fast that.
Didn't catch it until later on and spent most of an afternoon thinking
"why is it thinking it's linux?" and "why is it looking for gtk?"

Filed a bug on it.
https://bugzilla.mozilla.org/show_bug.cgi?id=468247

GDB debugging

2008-12-03T10:19:00.002-05:00

Currently I am getting acquainted with GDB.
There was a lab on debugging in Windows.

to debug an application using gdb

run a firefox debug(debug build of an app)

get the pid go to
ps aux | grep Minefield

call gdb

pass

full path of the executable
obj to firefox-bin -p (process ID)

ex
gdb /Users/usrname/mozilla/ffsrc/obj-ff/dist/MinefieldDebug.app/Contents/MacOS/firefox-bin 1266

note: 'warning' it might be an old build rebuild it again

commands
b breakpoint
b nameoffile:linenumber
c continue (frozen until continue)
l lists code around line
n executes next line of code
s follows code in function stepin
bt backtrace
quit

fprintf(stderr, "whatever %d", $var);
stderr

Hg Queues...

2008-11-24T10:28:00.002-05:00

references:
https://developer.mozilla.org/en/Mercurial_queues
http://zadkielm.blogspot.com/2008/11/hg-add-to-patches.html
http://hgbook.red-bean.com/hgbookch12.html

is a better way to manage your patches...it is a little bit complicated though.

hg revert (old way to "unimport" things)
hg queues are better.

for hg queues add this in your .hgrc

[extensions]
hgext.mq =

[diff]
git = 1

then in the repository that you want to use it type

hg qinit -c

some commands

hg qrefresh
hg qnew [patch-name] (create a new queued patch)
hg qpop [patch-name] (remove patch by name]
hg qdiff -U 8 > [patch-name]

Mozilla firstxpcom

2008-11-23T21:28:00.003-05:00

the lab

Repetition...it didn't work for the first few times.
I thought I wasn't following the steps right.

Had to run
make -f client.mk build
to get it to work

firstxpcom.zip

Interesting how it requires errors to create it...

Colour Profile Reftests on Virtual Machines

2008-11-19T09:26:00.004-05:00

Tests that run on Ubuntu 8.04/8.10 and Fedora 8 passes on Virtual Machines.
However if the tests are ran on non-virtual machines they fail.

Virtual Machines causes a level of weirdness...when it comes to testing
colour profiles.

Building Issues on my Mac OS X

2008-11-18T16:25:00.002-05:00

Started the Lab
then it told me to build.

and for some reason my Mac OS X build
is looking for a "gtk" package on linux.
the build thinks it is Linux....

I was told my that it was "goofy"...
Config Log
http://pastebin.com/f41e18f44
.mozconfig
http://pastebin.com/f4af56b69
error
http://pastebin.com/f59e527c4

My "new" hardware

Model Name: MacBook
Model Identifier: MacBook5,1
Processor Name: Intel Core 2 Duo
Processor Speed: 2.4 GHz
Number Of Processors: 1
Total Number Of Cores: 2
L2 Cache: 3 MB
Memory: 4 GB
Bus Speed: 1.07 GHz
Boot ROM Version: MB51.0073.B02
SMC Version: 1.32f6
Darwin 9.5.1

Colour Profile Reftest 0.2...and then there were two.

2008-11-14T20:36:00.008-05:00

Colour Profile Reftests on Firefox checks to see if the browser
displays the proper colours according to the colour
profile attached to it.

In this case there are:

Adobe1998
sRBG-IEC61966-2.1
ProPhotoRGB
ColorMatchRGB

This patch has the tests
Bug 463221

Each with
Red, Orange, Yellow, Green, Blue, Purple, Black, White, redpanda image
All of which have the 4 profiles attached to it.

Note: Patches cannot go over 1270kb (images included :( ).

Purposely failing a colour reftest will provide
access to the URI (the forced srgb image that reftest browser reftest renders).
This image URI needs to be cropped to the size of the original(with the profile)
and only then one can run the test comparing
the png with the profile against the image firefox renders.
(highlighted here)

note: having issues patches adding and the folders and files when applied.
created the patch, however when applied it doesn't include the folders and files.

ex. folder1
|-reftest.list
|-orange.list
|-imagesfolder(missing)

|-img.png(missing)

but the patch seems to have this information.

Ran failed pixels test
Bug 459617
on Mac Parallels
(however it seems to pass...it might be
because it is running in Parallels)

However running the test again
on a Fedora(F8) machine (Liberia)
shows that it still fails. Hurah...
note: the mac osx lets you remote into it ssh -CX
(compress and X11 forwarding...this was AMAZING!!)

editing ID3 tags in PHP

2008-11-09T22:42:00.005-05:00

found this set of libraries that does it.
http://code.google.com/p/php-reader/

downloaded it

and placed the 'src' folder in my Library

on a mac /private/etc/php.ini

include_path = ".:/php/includes:/Users/evmarbella/Sites/Library"

it requires memory to be
allocated to it in order to
run the script
or else you will get

Fatal error: Allowed memory size of 12582912 bytes exhausted (tried to allocate 1621950465 bytes) in /Users/evmarbella/Sites/Library/Reader.php on line 132

so in the php.ini
you need to find the line

memory_limit = 12M ; Maximum amount of memory a script may consume (12MB)

changed it to

memory_limit = 1600M ; Maximum amount of memory a script may consume (1600MB)

and it works.
as to why it requires 1600M not sure.

so I got it to read a mp3 file with
ID3v2.4.0

http://code.google.com/p/php-reader/wiki/ID3v2

FirePHP and phpunit

2008-11-09T14:30:00.006-05:00

references:
http://zadkielm.blogspot.com/2008/11/phpunit-mac.html

placed all the PHPUnit stuff in a Library folder
then I downloaded
FirePHP

to install it manually.
download here

unzip and place the FirePHPCore
in the Library folder
along with the phpunit framework.

to use it you have to include:

require_once 'FirePHPCore/FirePHP.class.php';
require_once 'FirePHPCore/fb.php';

along with

require_once 'PHPUnit/Framework.php';

to use it you need FireBug
I believe you need

Net

you can now send
logs

$firephp = FirePHP::getInstance(true);
$var = array('i'=>10, 'j'=>20);
$firephp->log($var, 'Iterators');

note: you might need to start output buffering

ob_start();

if it's not on in the php.ini

it should make it easier to peruse the PHP...

note: you do not want firephp to be outputting anything when you make your
site live.

Extensions Lab

2008-11-09T09:45:00.004-05:00

relatively the same steps...(Lab)

but for the mac there is this extension folder(not file) to put it in.

Users/username/Library/Application%20Support/Firefox/Profiles/eaf36vms.testfirefox/extensions/

note: id in the install.rdf and foldername should be the same.

some changes & notes:
http://mxr.mozilla.org/mozilla/source/browser/base/content/tabbrowser.xml
https://developer.mozilla.org/en/Setting_up_extension_development_environment

hg add to patches

2008-11-09T06:47:00.006-05:00

Adding a whole new test suite.

hg add filename or filepath

note: that it is recursive.

then hg diff...to a patch.

Patches cannot go over 1270kb
having images will make you go over
that amount.

note: having to remove things from the test now.

Messed around with the folder contents
removed the folder from the repository and added it back in again.

hg remove filename or folder

hg add filename or folder

hg import patchname to apply test suite

WAMP and PHPUnit

2008-11-05T16:41:00.008-05:00

For PHP on Windows install a WAMP server
http://www.wampserver.com/en/

then in command line

Start>run>cmd

go to

cd \wamp\bin\php\php*

note: whatever version of php* you have

run

go-pear.bat

it will ask a bunch of things (go with the default just to be safe)
then it will ask to change something in your php.ini file
you want to agree to those changes
once you have pear installed
in the same folder
run

pear channel-discover pear.phpunit.de

and

pear install phpunit/PHPUnit

this should install phpunit in the

\wamp\bin\php\php* folder

Now you will need to set your Environment Variables so that it finds
libraries you installed in your php* folder

go to Start > (right-click) My Computer > Properties > Environment Variables

and in

System Variable > Path

append this

;C:\wamp\bin\php\php5.2.6*

note: *php5.2.6 is the version running on this machine.

now you should have it all setup.

how to write test and whatnot.
http://www.phpunit.de/

PHP and PhpUnit Mac 10.5

2008-11-04T21:54:00.008-05:00

First 10.5 doesn't have php turned on.

Macintosh HD:private:etc:apache2:httpd.conf

cd /private/etc/apache2/
nano httpd.conf

locate

#LoadModule php5_module libexec/apache2/libphp5.so

remove '#'

LoadModule php5_module libexec/apache2/libphp5.so

then

cd /private/etc
sudo cp php.ini.default php.ini

note: a semicolon means it's commented out ;

locate

error_reporting = E_ALL & ~E_NOTICE

change to

error_reporting = E_ALL

to test

create a test.php with this

in it in your main web folder
ex.~/Sites

Installing PHPUnit Manually
some instructions here
http://www.phpunit.de/manual/3.3/en/installation.html

download the tarball from http://pear.phpunit.de/get/
extract it

(moved the 'PHPUnit' to ~/Sites/PHPUnit/)

go to

cd /private/etc
sudo nano php.ini

locate

include_path

should look like this

; UNIX: "/path1:/path2"
include_path = ".:/php/includes:/Users/evmarbella/Sites/PHPUnit"

note: the :/Users/evmarbella/Sites/PHPUnit
is the the PHPUnit folder

save it

locate a file called pear-phpunit
that was extracted from the PHPUnit tar.

rename it

mv pear-phpunit to phpunit

then

nano phpunit

locate

@php_bin@

replace with

/usr/bin/php

(this is where php is)

save the file

do a

$PATH
-bash: /opt/local/bin:/opt/local/sbin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin:/usr/X11/bin: No such file or directory

move the file to /usr/local/bin

mv phpunit /usr/local/bin/

now

cd /usr/local/bin/

then

chmod +x phpunit

then

cd ~/Sites/PHPUnit/Util/
nano Fileloader.php

replace @php_bin@

with

/usr/bin/php

restart the WebSharing from
Apple>System Preferences>Sharing>Web Sharing

then you should be good to go*

It was working then error:

Warning: require_once(PHPUnit/Util/
Filter.php): failed to open stream: Not a directory in /usr/local/bin/phpunit on line 40

Fatal error: require_once(): Failed opening required 'PHPUnit/Util/Filter.php' (include_path='.:/Users/evmarbella/PEAR') in /usr/local/bin/phpunit on line 40

it changed to

Warning: require_once(PHPUnit/Util/Filter.php): failed to open stream: Not a directory in /usr/local/bin/phpunit on line 40

Fatal error: require_once(): Failed opening required 'PHPUnit/Util/Filter.php' (include_path='.:/php/includes:/Users/evmarbella/Sites/PHPUnit') in /usr/local/bin/phpunit on line 40

note: it was looking in the wrong folder. ':/Users/evmarbella/Sites/'

dive into modifying firefox lab

2008-11-03T09:09:00.005-05:00

This is where I found the tab.
http://mxr.mozilla.org/seamonkey/source/xpfe/global/resources/content/bindings/tabbrowser.xml#1011

search for attributes of mTabContainer

760 var listener = this.mTabListeners[this.mTabContainer.selectedIndex];

where I found it
browser/base/content

in tabbrowser.xml
1126

// Insert tab after current tab, not at end.
if (this.mTabContainer.childNodes.length == 1) {
this.mTabContainer.appendChild(t);
} else {
var currentTabIndex = this.mTabContainer.selectedIndex;
this.mTabContainer.insertBefore(t, this.mTabContainer.childNodes.item(currentTabIndex + 1));
}

1234 // wire up a progress listener for the new browser object.
var position = currentTabIndex + 1;

obj-ff/dist/MinefieldDebug.app/Contents/MacOS/firefox --no-remote -P

being immersed in colour testing kinda removes me
from this kinda thing. the tools I use are like
gimp and the reftest platform.

Had some issues with the tabs closing.
But apparently others had the same problem.

reftesting other configurations

2008-10-19T16:30:00.005-04:00

So I started a contributions page on my wiki.
I did the first one...just to see.

So I spent a morning and a better part of an afternoon
installing Ubuntu on my Mac OS X 10.4.11 with Parallels
and ran the test and it passed. (it's suppose to fail)

Note:
I was told that running things in a virtual machine
might be messing around with the tests...
I'm looking to see if others are failing also.

My first test was on Ubuntu but it was
installed using Wubi.

FSOSS...can't wait.

2008-10-19T14:57:00.009-04:00

FSOSS is coming soon and there is a stream of talks
on Teaching Open Source or TOS @ FSOSS (pretty catchy).

When someone is trying to teach me something new
I always think of

how well they are making the hippo dance?

http://blogs.wnyc.org/radiolab/2008/09/09/making-the-hippo-dance/

The reftest and I v0.1

2008-10-12T22:22:00.005-04:00

So after a few weeks getting to know
how to do reftests with colour profiles.

a chronicle of what has occurred so far...
http://zadkielm.blogspot.com/search/label/reftest
note: the blog is really beneficial for reference.

I have a test that fails. Which is good news!
It means there is direction a place to start
digging and discover new things.

Filed a bug.
https://bugzilla.mozilla.org/show_bug.cgi?id=459617

The good:
-Learning so many things.
-Reading more.
-It's always changing.

The bad...well it was ok:
-It's always changing.
-So little time.
-Starting to spell colour 'color'

I really appreciate all the help
and direction I've received so far...

Pixel reftesting with profiles - Isolating the pixel - replicating failures

2008-10-11T12:23:00.011-04:00

So apparently I sampled the wrong image and
shuffled the steps/files...it gets confusing

replicating
steps 1-7 done on either a mac or windows
step 8 on ubuntu

sample the failing pixel from the original image
create an image with just that pixel
apply the profile to it(make sure that its the same one profile of the test that failed)
create a failed test
gather the failed test URI output and
paste it in the address bar of the firefox.
crop and save the firefox rendered png
make a test with the pixel with the profile
and the firefox rendered reftest image

It fails again on the Ubuntu.

Pixel with profile Adobe 1998
8ca39b = ffrender 82a49c (on a Mac*) fails on the Ubuntu
6f8753 = ffrender 63884f (on a Mac*) fails on the Ubuntu

note*: it's probably the same on windows. They are interchangeable.

83a49c = reftest ubuntu pixel
64884f = reftest ubuntu pixel
r-g-b-

update:
it is off-by-one in the red channel
Will need to try the tests on other configurations...this might mean some contributions.

When a Fail turns to a Pass.

2008-10-10T22:24:00.007-04:00

What do you do when the problematic reftest pixels are passing?
well then you sit back down and rethink if you have
sampled the right pixel...or maybe it's the pixel with
the colour profile you should be sampling...or...or...or...

Failed Pixels (Adobe 1998) from redpanda*.png
Sampled Colors
8ca39b
6f8753

reftesting - issues uncovered...

2008-10-06T09:24:00.006-04:00

notes from : emails

so after a few weeks of getting things right
finally uncovered some issues that can be
investigated.

After talking to bholley on the irc and through emails.
I was told that the sRGB is working for Ubuntu and that
I might have uncovered something else.

So I will look in to whatever that "something else" might
be.

which means taking a failed pixel and seeing why it's failing.

Img A: windows and ubuntu layer difference showing which pixels are failing.

Img B: mac and ubuntu layer difference showing which pixels are failing.

debugging

2008-10-03T23:37:00.003-04:00

venkman and gdb...seems to be the way to go.

Reftesting Colour Profiles and the ubuntu

2008-10-03T22:53:00.005-04:00

My current reftest seems to proving
that the Ubuntu firefox doesn't
seem force sRGB during reftests:

based on redpandatests.list
(which is a series of redpanda pictures with color profiles.)

If tests pass it means the images were outputted in srgb.

Mac test with Macrenders = pass
Mac test with Ubunturenders = fail all but passes sRGB redpanda_ICC_sRBG-IEC61966-2.1.png
Mad test with Windowsrenders = pass

Ubuntu test with Macrenders = fail all but passes sRGB redpanda_ICC_sRBG-IEC61966-2.1.png
Ubuntu test with Ubunturenders = pass
Ubuntu tests with Windowsrenders = fail all but passes sRGB redpanda_ICC_sRBG-IEC61966-2.1.png

Windows test with Macrenders = pass
Windows test with Ubunturenders = fail all but passes sRGB redpanda_ICC_sRBG-IEC61966-2.1.png
Windows test with Windowsrenders = pass

I think, I can conclude that Ubuntu is not forcing sRGB during reftests.
So where do I go from here?

Firefox 3.1 Build Using VS express 2008

2008-09-30T19:26:00.003-04:00

notes from: http://zenit.senecac.on.ca/wiki/index.php/User:Jamesboston

note: my goodness do things ever take sooo long to download and install...Downloading

VS Express C++ edition

note: I wonder if you can add more editions to it...i want C# on there also.

Then you gotta register it
Visual Studio Express 2008 Registration Key

Download
Microsoft® Windows® Software Development Kit Update for Windows Vista™
you need WGA

Download
[ ] MozillaBuildSetup-1.3.exe or higher
http://ftp.mozilla.org/pub/mozilla.org/mozilla/libraries/win32/

install Microsoft ® Windows Server® 2003 R2 Platform SDK Web Install

* You need to install at least the "Windows Core SDK", "Web Workshop (IE) SDK", and the "Data Access Services (MDAC) SDK".
* For all SDKs, you may choose not to install Sample Code and Documentation to save space and download time

did a "typical" install.

Install all these 4

yay it all installed with no problems

using experimental mozilla-build\start-msvc9.bat

cd
mkdir mozilla
hg clone http://hg.mozilla.org/mozilla-central/ src

mozconfig

# Mozilla config
. $topsrcdir/browser/config/mozconfig
mk_add_options MOZ_OBJDIR=@TOPSRCDIR@/obj-@CONFIG_GUESS@
mk_add_options MOZ_MAKE_FLAGS="-s -j4"
ac_add_options --disable-optimize
ac_add_options --enable-debug
ac_add_options --disable-static --enable-shared
ac_add_options --enable-tests
#vs2008
ac_add_options --disable-xpconnect-idispatch
ac_add_options --disable-activex
ac_add_options --disable-activex-scripting
ac_add_options --disable-accessibility
ac_add_options --disable-parental-controls
#XP Build
ac_add_options --disable-vista-sdk-requirements

note: had to move the folder to C:\mozilla\src
because document and Settings has spaces...it didn't like that

make -f client.mk build

it's building...it takes a while.