掲示板 過去ログ
(2013年1月1日 〜 2013年12月31日)

kuma-taさん初めまして。
現在、当サイトで公開しているビルドは32ビット(x86)版のみです。64ビット(x64)版のtmemutil.dllを、32ビット版のビルドに入れて使うことはできません。
x64版のtmemudil.dllは、他の方のビルドで使われる場合があるかもしれないということで公開してます。

横から失礼します。

テテさんの作成されたThunderbirdをお使いということならば，“tmemutil-20120427-sse.zip (10KB)”をお使いになられればよいかと思います。
または公式ビルドのESR版24.2.0をお使いということであれば，同様に上記のものがお使いいただけるかとは思いますが，効果のほどはいかほどかは不明です。

もしも後者ということであれば，テテさんの先の言からもテテさん作の私的ビルドではなくとも何某かの効果は期待しうるという意味では，お使いいただけるということで間違いありません

64ビット(x64)版のtmemutil.dllを、32ビット版のビルドに入れて使うことはできないのですね。
勉強になりました ありがとうございます。

koyamaさん、初めまして。
Firefoxでwebページに埋め込まれたMIDIファイルを再生するには、QuickTime等のアドオンをインストールする必要があるようです。

webページにMIDIファイルを埋め込む方法は、私の知っている限り数種類ありますが、body要素のbgsound属性(非標準)で埋め込む方法は、Firefox等は非対応のようです。
https://developer.mozilla.org/ja/docs/Web/HTML/Element/bgsound

Firefoxの場合、embedやobject要素でMIDIファイルがwebページに埋め込まれているなら、QuickTime等がインストールされていれば演奏できるのではないでしょうか。

I have uploaded it. I hope this version will be the final 17.x.

Ok great...thanks once again.

For me 25x & 24x are very buggy (slow and don't load pages properly etc) so 17x is like a breath of fresh air...I can't fault it.

I hope they keep it going whilst there are so many problems with the current version anyway...we'll see.

Cheers Tete.

Hi, I have just uploaded my build of Firefox 17.0.10 ESR as a memory.

Thank you, Tete. Maybe you'll find some time to build TB 17.0.10 as well

Thank you very much Tete, version 17 is still flying

I will never use 24x again

Sorry I never responded earlier...I can't connect to this board easily...it takes ages from the UK...I think it took twenty attempts before I could get in...maybe it's busy but I don't know.

Anyway I'm happy now...thanks again.

Sayanora

I'm glad you enjoyed it! I uploaded my build of Earlybird 17.0.10.

Great! Thanks again, Tete.

おはようございます。
昨夜、Firefox 25.0独自ビルドのビルドに失敗してやり直していたのですが、途中で眠ってしまったため、今続きをやっているところです。
問題が起きなければ、今晩アップロードしたいと思います。

失敗の原因は、xul.dllをリンクしている時にリンカのメモリ使用量が増えて、アドレス空間の壁(4GB)に達したためと思われます。

お疲れ様でした。
さっそく頂戴しました

どういたしまして。(^^

To tell the truth, I don't like Australis much, but if it will be formally adopted in the future, I am not going to disable it.

It would be really nice and much appreciated if you then could release two versions (one with and one without Australis)! :)

Regards.

ヒロヒトさん、こんにちは。今のところ私のところには、同様の不具合の報告は寄せられておりません。

公式Firefox 23.0.1では、その不具合は発生しませんか?
また、ご使用のWindowsの種類を教えていただけると、何かの参考になるかもしれません (例: Windows 7 64ビット)。

Thank you for reporting the problems. I have confirmed the problems.
I think some illegal code was generated by Profile-Guided Optimization (PGO) or -GL compiler option with MSVC2005.
It may take a long time to narrow down the place of the problem and fix it.

It's okay. Take your time no need to rush

I (and maybe other users of your build) can run both your previous build (FF22) and the new FF23 build from Mozilla side by side if it's really needed in the meantime*

* for example to experience the new DXVA feature in FF23 lol

I have updated my build of Firefox 23.0.
As a stopgap measure, I replaced a xul.dll by the non-PGO version.

Thanks for the update. Hopefully you can find a fix so it'll be possible to use PGO on xul

I have updated a new version with a bug fix.
It seems that the problem was fixed by excluding content/media/AudioStream.cpp from PGO.

Just checked now (because i was outside yesterday) and now videos run normally. Thanks for the fix, that was fast!

どういたしまして、と言いたいところですけど、mp4動画の再生中に映像が止まるという不具合を確認しました。
PGOか-GLオプションの不具合に当たった可能性が高そうです。というのも、同一ソースからビルドしたデバッグ用のxul.dllに差し替えたら、不具合が治まりましたので。

というわけで、不具合の発生箇所を絞り込むのに、かなり時間がかかりそうです。

http://www.youtube.com/watch?v=jOBuQ_-UwlU&html5=1

この動画だと音声は正常、
動画は2倍速？になって再生されたりします。

もうご存じかも知れませんが、
問題解決の糸口になればと思い報告させて頂きます。

通りすがりさん、情報ありがとうございます。参考にさせていただきます。

動画が正常に再生されないのは、WebMだけでなく、WebMやMP4もでした。

間違えました。WebMだけでなく、TheoraやMP4もでした。

Firefox 23.0 独自ビルドの不具合修正版をアップロードしました。

content/media/AudioStream.cppをPGOから除外したら、動画再生時に映像が停止する不具合は治まったようです。
不具合をご報告くださり、ありがとうございました。

I changed the value of that option in my build of Firefox 22.0 and restarted the build, but the value was not reset. I tested it in new profile.

I tested it with a new profile now. You are right. Seems my profile is broken and I have to make a new one.
Thank you for your help!;-)

you can download old version at

skydrive tete009
https://skydrive.live.com/redir?resid=A6D7F1464F4932AC!110

既にnatさんが一つ紹介してくださいましたが、現在、以下のオンラインストレージに、今までのビルドの一部をアップロードしています。

Google Drive: https://docs.google.com/folder/d/0BwJVYWis62cRYW9PSHFqcDk5S2M/edit?pli=1
SkyDrive: https://skydrive.live.com/?cid=a6d7f1464f4932ac&id=A6D7F1464F4932AC!110

natさん、テテさん
ありがとう！

I'm new to your builds and was also looking for old versions.
Thank you so much!
natさん、テテさん
ありがとう！

I have no thought of making such a build, but I suggest you adjust the value of layout.css.devPixelsPerPx.

I've already done, but FF font is still smaller than previous version :(

@Firefoxer you can switch back to v21 all time.

I don't know whether this problem is inherent in my build, but I found news articles that D630 may experience the NVIDIA GPU issue over time.
http://en.community.dell.com/dell-blogs/direct2dell/b/direct2dell/archive/2010/09/30/nvidia-gpu-update-nvidia-class-action-lawsuit-and-limited-warranty-enhancement.aspx

If you have not updated the BIOS of your D630 to the latest one, I think you should update it.

Thanks for the link. I was not aware that nvidia in d630 has an issue. I was just wondering if the crash was related to your build (which is otherwise great). It was just one video so no big deal. I have refrained from updating bios since I had the motherboard replacement before (under warranty, not booting up). I was regularly updating bios before that. Thanks.

@ryan you has try reproduce with offical firefox build,or pcxbuild,lawliet build? has same issue?
i test with my samsung r478 with tete build v21 no issue (not a dell )

@RyanV

And btw, did your youtube use flash or html5? And after close HW feature, the crash will be reproduced?

Sorry for the late response. I haven't had the chance to try it with the official build. And I think that particular video was html5 (whatever that means). Nevertheless, I upgraded to the recent v22 tete build and tried the video again - - - no more problems or crashes. I guess it was something with that particular (v21) build. ;)

I remembered I wrote the details:

http://cgi38.plala.or.jp/tete009/board.cgi?id=T1334220309.89#T1334220309.89

(2) and (5) has been removed.

Thank you xunxun. The features of my builds are just as xunxun wrote.

I think it has been pretty much the same patches in my Firefox18/19/20 builds.

i think next version v22 tete will use vs2010 as mozilla stop
compat vs2005:-D

Hi nat, I probably won't be able to meet your expectations for the time being.

hi
refer from https://bugzilla.mozilla.org/show_bug.cgi?id=866425
"Since VC10 is now available for no charge"
i don't know where we can get vc10 for free and use instead current.
but may be fyi for tete

regards
nat
tete009 user.

I know that bug.
At the moment, I can build mozilla-central source code with vc8 and my own patches, but I don't know until when I can continue this.

Visual Studio Express edition (free) does not support PGO.

@nat

tete009 may revert some patches to solve the problem, as long as Mozilla won't change the code a lot. At present, the greatest impact for tete009 is that Metro UI will land in the release some months later (VC2005 and Win8 SDK's compatibility is unknown).

@tete009

You should consider to use Paypal.

I also want to update my VC toolchain to VC2012, but no money to buy it Hope Intel send me a free-used compiler.

At the past, I want to build Mozilla code using free MinGW(64) GCC, but that will break much more compatibility.

@xunxun
I have no plan to use PayPal at present. If I require new version of Visual Studio, I would like to buy it

i test pcx firefox and tete build. but i confuse that
why pcxfirefox icon is same official build but tete009 is nightly icon? both pcx and tete is use release code channel? please explain as i can easy understand. i am normal user

@nat
The effect depends on whether builder specify - -enable-official-branding option at build time.

may have other compiler that use instead of visual studio? i not know what chromium browser compiler on windows, such as coolnovo,torch,srware iron,maxthon4
but would has free compiler for windows. and intel complier is free or must buy to use?

I don't know a lot what those software are compiled with. The Windows version of Intel C++ compiler is not free.

I suggest not to build Firefox using Intel Compiler on Windows, except you have enough time to solve many bugs or only for some special purpose.

firefox 22 tete build work smooth and stable fastest than official build thank's for your coding.:-D

Hello tico-tico.

Can you send me your minidump file?

If CreateCrashDump entry in the [General] section of tmemutil.ini is set to 1, my builds may be able to generate a minidump file.
http://www1.plala.or.jp/tete009/software/mozilla/tmemutil.ini

My email address: tete009@gmail.com

@tete009

Well, you change your Email ....

@tico-tico
Thank you for sending me your minidump file.
According to callstack, crash signature seems to be the following:

Crash Reports for nsCOMPtr_base::assign_with_AddRef(nsISupports*) |
nsSocketTransport::SendStatus(unsigned int)
https://crash-stats.mozilla.com/report/list?product=Thunderbird&query_search=signature&query_type=exact&query=nsCOMPtr_base%3A%3Aassign_with_AddRef%28nsISupports*%29%20|%20nsSocketTransport%3A%3ASendStatus%28unsigned%20int%29&reason_type=contains&date=05%2F06%2F2013%2014%3A38%3A44&range_value=1&range_unit=weeks&hang_type=any&process_type=any&do_query=1&signature=nsCOMPtr_base%3A%3Aassign_with_AddRef%28nsISupports*%29%20|%20nsSocketTransport%3A%3ASendStatus%28unsigned%20int%29

Bug for nsCOMPtr_base::assign_with_AddRef(nsISupports*) |
nsSocketTransport::SendStatus(unsigned int)
https://bugzilla.mozilla.org/show_bug.cgi?id=565638

I don't know what the cause of the crash is...

@xunxun
I have several email addresses.

Oops, the link of crash report URL was not generated correctly. Pastes the shortened URL:

Crash Reports for nsCOMPtr_base::assign_with_AddRef(nsISupports*) |
nsSocketTransport::SendStatus(unsigned int)
http://bit.ly/10gNMp7

There are many more crash reports for Thunderbird 17.0.5 than for Thunderbird 17.0.5esr.

Crash reports for nsCOMPtr_base::assign_with_AddRef(nsISupports*) | nsSocketTransport::SendStatus(unsigned int)

Thunderbird 17.0.5
http://bit.ly/10f0Sxl

Thunderbird 17.0.5esr
http://bit.ly/127WpAL

I noticed TB 17.0.4 has not the crash ID, maybe some 17.0.4-17.0.5 changes cause that.

@tete009

You can refer to these changes:

Security Changes: http://www.mozilla.org/security/known-vulnerabilities/thunderbird.html#thunderbird17.0.5

Bugfix Changes: https://bugzilla.mozilla.org/buglist.cgi?quicksearch=ALL%20tracking_thunderbird_esr17%3A%2220%2B%22%20%20status_thunderbird_esr17%3Afixed&list_id=6480427

@xunxun
It appears this crash is also reported by older versions of Thunderbird.

I've been trying to reproduce the crash in the official version (with IMAP-acc only) for three days. Still no "luck".

Thank you for reporting the problem.

I have not been able to recreate the problem you told me about. I use my builds on Windows 7 x64 and Intel HD Graphics 3000.

Can you tell me an URL where the problem occurs easily? If I can recreate the problem on my system, it may be able to solve the problem.

Thanks you for the reply.
I can easily reproduce bug scrolling fast up-down on this page: http://webspoon.ru/receipt/snek

Well, it's not so easy to reproduce but I managed to record video http://youtu.be/AayAPif7WXU

Sorry for triple post.
There's definitely something wrong with GPU acceleration. I tried to set it to 4-th (out of 6) position. It's something like "Disable all parameters of acceleration for cursor & complex images rendering" (not exact phrase, I translated). The problem disappeared.
Then I tried to change old driver with mixed modules (used it to improve performance of some old games) to latest official. This bug didn't happen so far...

Continued experiments. I knew that performance hit (e.g. from 59 to 29 fps) of newer AMD drivers was caused mostly by 2 modules: ati3duag.dll (Radeon (R300 Family) Windows 2000 DirectDraw/Direct3D Universal Driver) & atikvmag.dll (Virtual Command And Memory Manager). So I had to replace them with older versions. After that problem appeared again :/
I can only guess that this bug related to fast rendering. I doubt that you'll be able to reproduce it (different OS, different GPU) but still there's possibility that disabled vsync may 'help' you to do it.

Thank you for all the information.
I had tried to reproduce the problem, but I could not reproduce the problem on my system.
I have not been able to find out what has caused that...

Thanks for trying!
Anyway, it's not such a big problem with 'newdisp' http://msdn.microsoft.com/en-us/library/windows/hardware/ff568527%28v=vs.85%29.aspx
It's not so easy to find so here's few links for those who needs it:
http://www.mediafire.com/?m3shjt1s6r2xpi3
http://cid-a50350ea7a969f0c.office.live.com/self.aspx/Public/newdisp.exe

Maybe it's not caused by tete009 multithread feature patch, because I introduce the feature to pcxfirefox for a long time since Firefox 12.

But there's not only multithread feature patch
Now I have this artifacts in social network 'VK'
http://imgur.com/eDFO4VJ
For now I just disabled buggy hardware acceleration because driver change isn't an option for me.
Edit > Preferences > Advanced > General > Browsing: "Use hardware acceleration when available"

Ah, there definitely should be 'edit' button
Just tried 17th version. At the first blush there's no artifacts on my system. Seems like I shouldn't change it & wait for Mozilla's fixes.

yea teteh is the man :) sugestion
starting from this version 20 maybe think about a 64bit version

Today the 64 bit versions with enhanced code firefox that would be faster than the 32bit version? or not yet? teteh

Tete won't build Fx20 x64 this time.

somtetime i scan pcxfirefox in virustotal.com and show as has some trojan?. please tell me why?

I have uploaded my build of Firefox 20.0, etc.

Also refer to https://bugzilla.mozilla.org/show_bug.cgi?id=653342 Possible fix, please.

Untested unlanded unreviewed WIP patch? Are you sure it's a good idea?

grrr... it was me.

I have tested it, and it works well. I don't know why it's not reviewed.

i think it's because the patch author didn't answer to the reviewer question (https://bugzilla.mozilla.org/show_bug.cgi?id=653342#c105) and seems like completely disappeared. i'm not saying the patch is bad but i'm not sure about possible regressions.
anyway the decision is up to Tete.

I applied the patch of Bug 493544 to my Earlybird 17.0.4.

Well, it is often burdensome to me to track the patches of development versions. So I don't want to apply the patches of development version as much as possible.

I would appreciate your understanding this is a special case.

Thanks for the new build Tete. Everything seems to work fine.

Out of curiosity, why you don't use optimization for the omni.ja files? Tb and Fx should start faster with the optimized omni.ja's, at least Mozilla says so.
I optimize them for every your builds but to be honest can't see any improvements. But maybe it's just my case.

What do you mean by "optimization for the omni.ja" ? Compression?
It seems that omni.ja files of my builds were already quite compressed.

omni.ja's shipped with the official version are not regular zip-files. if you open such file you'll see that it is not started with "PK" symbols. mozilla uses the optimizejars.py script (can be found somewhere in the sources) to get fast-loading version.
i can't find anything about it on MDN, maybe this helps https://blog.mozilla.org/tglek/2010/09/14/firefox-4-jar-jar-jar/ ("Optimization #3: Optimized jar layout" section).

Thank you for the info.

I guess that the optimization information for JAR files is collected when collecting the profile data for PGO. But I didn't set a specific environment variable which seems to be needed for JAR file optimization.

I don't thing that jar-optimization related to PGO. Coz i optimize it without any PGO info, i even don't have compiler with PGO support just Express edition for my home needs.
I can send you compiled (if we can say "compiled" about python scripts, but it's an exe-file) optimizejars.py script for testing.

I found the Bug 559961 - Reorder jar files as part of PGO
https://bugzilla.mozilla.org/show_bug.cgi?id=559961

When we set the MOZ_JAR_LOG_DIR environment variable and run Firefox 19, etc., the build saves jar log files into the specified directory.

I guess we can optimize the layout of jar files by passing the directory as an argument to optimizejars.py.

The optimizejars.py is in the "config" directory of source code.

@tete009

I think the process is effective only when using make installer? How is the process related with PGO?

if you run the optimizejars.py script (Python 2.7 required) you'll see

>Usage: - -optimize|- -deoptimize JAR_LOG_DIR IN_JAR_DIR OUT_JAR_DIR

not a sign of PGO

I think jar log files collection is done automatically when collecting the profile data for PGO, if builders specify profileserver.py to PROFILE_GEN_SCRIPT.

This is only my own guess, but jar optimization using the log files might be done when builders run "make installer".

Oups, it seems like what i was doing (optimization without log files) didn't make much sense. Great you figured it out.

Thank tico-tico to point that.
Generally I omit the jar optimization line.
I will re-enable it on my next build.

Well, I try to use the method, but nothing generated is in my jarlog directory. I don't know why.

@xunxun
Since mozilla21, the MOZ_JAR_LOG_DIR environment variable is not used.

Instead, we need specify the output path of a jarlog file to the MOZ_JAR_LOG_FILE environment variable when we don't use official profileserver.py. And when we build en-US build, we need to put the jarlog file to the path of "($OBJDIR)/jarlog/en-US.log".

As you know, old versions have many security risks. I recommend you try to use the add-ons such as Disable Add-on Compatibility Checks.

You can download my old versions from the following online data storage services.

Google Drive: https://docs.google.com/folder/d/0BwJVYWis62cRYW9PSHFqcDk5S2M/edit
SkyDrive: https://skydrive.live.com/?cid=a6d7f1464f4932ac&id=A6D7F1464F4932AC!110

Thanks a lot for the links and the heads up!

Problem solved.I replaced the rest files with Resource Hacker.

I'm glad to hear that you found a way to resolve the problem.

No longer using plala.or.jp host ?

I can not access URLs you posted, but the only sites where I upload my builds now are plala, Google Drive and SkyDrive. I will continue to use plala.

I uploaded my builds of Firefox 19.0.2 and Earlybird 17.0.4.
But I'm rebuilding my build of Firefox 17.0.4 because of the fatal error of compiler.

Thanks

I'm glad you liked it. Thank you!

@Tete : Just let you know : tete Firefox 19 rocks !!!

@Firefoxer
It's my pleasure. Thank you!

I have just uploaded my build.

Hello Victor,

I gave up supporting Windows 2000 on my builds because Firefox 13 or higher uses many functions which don't exist in Windows 2000.

I have never tried Cyberfox so I don't know it...

Have you tried VS 2012 ?

No, I have not tried builing with VS2012 yet. I am still using VS2005 SP1.

While attempting to compile with the newest compilers is nice idea, its weird to announce that browser as being fast, since its even slower than original firefox in many tests, at least on my laptop. Cyberfox still need more tuning.

After comparing above builds with tete build, i found tete's one is still ahead and scores better than original FF18. One test where tete's ff is slower than ff18 is Sunspider.

Ups, my statement about Cyberfox need verification and more tests. Cyberfox is 64bit browser so no compare to 32 bit builds.

may be tete would change icon own build firefox.
to better looking if not use nightly icon.8-)

I will use the default icons for unofficial build, to avoid making troubles.

tete has plan to use vs2012? How difference with vs2005
as using in present?

I will continue to use vs2005 for a while. Professional edition, etc. are expensive for me.

thank you for upload firefox21. rapid i will use as my default browser. i to be tete fanclub. i live in thailand.;-)

In most cases, I build the source code of releases or candidates.
My unofficial builds are built without the "- -enable-official-branding" flag. Currently, unofficial release builds use "nightly" branding until Bug 659568 is fixed.

Bug 731043 - Builds from mozilla-beta shouldn't default to official branding
https://bugzilla.mozilla.org/show_bug.cgi?id=731043

Bug 659568 - Make "browser/branding/unofficial" the default for Firefox unofficial builds
https://bugzilla.mozilla.org/show_bug.cgi?id=659568

Thanks for the reply but just for clarification, would you say that your build is more stable than the beta builds that Mozilla issues for Firefox prior to the official release of FF?

In other words, would your build have the same amount of bugs or possible instability like the Mozilla Firefox pre-release builds?

My builds are usually built based on the source code of Release channel.

If there are not any specific problems of my builds, the stability is the same as the release version of Mozilla Firefox.

So would it be correct to say that you modify the source code of the official release builds of Mozilla Firefox to conform to your own optimized specifications?

It is not the Beta builds which you use to modify source code?

Finally, would it be correct to say that your build is as stable as the official Firefox release build 18.01 but only better because you optimize it?

Thanks.

My builds are usually built based on the source code of official Release, so my builds don't usually have the bugs which were fixed in development phase such as Nightly, Aurora and Beta.

But my builds differ in several ways from official builds, so I cannot say that the stability of my builds are strictly the same as official release builds.

* There may be a bug in my specific patches or DLL.
* Some kind of problems might be caused by a difference in the compiler version or options.
* Illegal codes may be generated according to the way I collect the profile data for PGO. Unfortunately this do happen occasionally.

Thanks for the clarification. Although I have used the official FF release, I find yours to be more compatible with my add-ons than your competitor Palemoon plus it operates faster. Moreover, you always seem to stay current with the latest release.

Keep up the good work.

One suggestion, why don't you give your browser a name instead of just the word "Nightly"? For example the Tete browser or Tristar Firefox? I say this because the word Nightly to me implied Beta which prompted all my questions, and all of which you gratiously answered.

I don't like "Nightly" branding very much for my release builds, but I reluctantly accept "Nightly" branding because in addition to new name, I need prepare new icons and images...
Thanks.

https://bugzilla.mozilla.org/show_bug.cgi?id=794246
https://bugzilla.mozilla.org/show_bug.cgi?id=807581
https://bugzilla.mozilla.org/show_bug.cgi?id=820247

There's also this bug. Maybe this is related. I had to go back to 17.0.2.

Thank you for reporting the problem.
Your minidump.dmp indicates that an access violation occurred in igd10umd32.dll. This is a file for Intel graphics.

I use Intel HD Graphics 3000 and driver 9.17.10.2932, too. But I have not been able to reproduce the problem. Can you tell me the way to reproduce the crash?

I would appreciate it if you could check whether the same problem occur with an official Firefox 18.

OOPS！ same problem occur with official firefox 18, all problems!
my laptop has intel graphics hd 3000(igpu) and nvidia gt520m(dgpu). in nvidia control panel, i can set firefox runs on dgpu or igpu. the crash can be reproduced while firefox runs on dgpu, but firefox will not crash on igpu.

Thank you for letting me know the more detailed information.

I wonder why some Direct3D function calls the function of Intel's igd10umd32.dll, even though you configure NVIDIA Control Panel to use dGPU on Firefox 18.0.

When you have configured NVIDIA Control Panel to use dGPU on Firefox 18.0, can you tell me Firefox's Graphics information in about:support ?

information of using dgpu(hardware acceleration enabled):
Graphics
Adapter Description Intel(R) HD Graphics 3000
Adapter Description (GPU #2) NVIDIA GeForce GT 520M
Adapter Drivers igdumd32 igd10umd32 igd10umd32
Adapter Drivers (GPU #2) nvd3dum nvwgf2um,nvwgf2um
Adapter RAM Unknown
Adapter RAM (GPU #2) 1024
ClearType Parameters Gamma: 2200 Pixel Structure: RGB ClearType Level: 100 Enhanced Contrast: 100
Device ID 0x0116
Device ID (GPU #2) 0x1050
Direct2D Enabled true
DirectWrite Enabled true (6.2.9200.16440)
Driver Date 12-12-2012
Driver Date (GPU #2) 12-29-2012
Driver Version 9.17.10.2932
Driver Version (GPU #2) 9.18.13.1090
GPU #2 Active false
GPU Accelerated Windows 1/1 Direct3D 10
Vendor ID 0x8086
Vendor ID (GPU #2) 0x10de
WebGL Renderer Google Inc. - - ANGLE (Intel(R) HD Graphics 3000)
AzureCanvasBackend direct2d
AzureContentBackend direct2d
AzureFallbackCanvasBackend cairo

Thank you for all the information.

You reportedly set up NVIDIA Optimus to use dGPU on Firefox 18.0, but according to the information of Graphics of Firefox, it seems that Firefox uses Intel HD Graphics 3000 for hardware acceleration. It looks strange to me that the item of "WebGL Renderer" contains the string "Intel(R) HD Graphics 3000".

If we see the crash reports which have the crashing signatures which contain the string "igd10umd32.dll", there is a sprinkling of the crashes apparently connected with NVIDIA Optimus.
https://crash-stats.mozilla.com/topcrasher/byos/Firefox/18.0/Windows/7/browser

This issue seems to have been reported already in bugzilla. I wrote a comment to call attention to this problem.

Bug 763585 - Firefox crash @ igd10umd32 with NVIDIA/Intel GPUs
https://bugzilla.mozilla.org/show_bug.cgi?id=763585

ff18&18.0.1 works fine on win8 x64.......... no crash =-O

You can follow Tete's twitter : https://twitter.com/tete009_

I have just uploaded my builds. I do not plan to build 10.0.12.

Please update 10.0.12esr because I have an old PC.:'(

I no longer build 10.0.12esr. Official old versions are available from the following site, but old versions have security risks.
https://ftp.mozilla.org/pub/mozilla.org/firefox/releases/