Jump to content

roytam1

Member
 View Profile  See their activity

  • Posts

    2,307

  • Joined

  • Last visited

  • Days Won

    326

  • Donations

    $0.00 

  • Country

    Hong Kong

 Content Type 

Posts posted by roytam1

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New regular/weekly KM-Goanna release:
    https://o.rthost.win/kmeleon/KM76.4.6-Goanna-20220528.7z

    Changelog:

    Out-of-tree changes:
    * update Goanna3 to git e7535dc94b...0bc2879198:
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1134252 - Don't crash the content process if RenderFrameParent is not constructed successfully. r=billm. (2564cb0e6a)
     - Bug 1180644: Fix crashes after enabling OOP on B2GDroid. r=snorp (d585c571e3)
     - Bug 1198674 - null-check mFrameLoader in RenderFrameParent. r=sotaro (86f26b2046)
     - Bug 1198674 - Null-check mFrameLoader before calling GetFrom in RenderFrameParent. r=sotaro (33bd495e75)
     - Bug 1200778 - Make sure to update the APZCTreeManager associated with a RenderFrameParent when it is dragged to a new window. r=mstange (bf2d25616c)
     - Bug 1202703 - Part 1 - CreateRenderingContext can fail. r=mattwoodrow (50de4cd050)
     - Bug 1185747 part 1 - Use pref/meta-viewport tag instead of DOMWindowUtils to set the CSS viewport for mochitests. r=tn (62a8c1d460)
     - Bug 1147038 - Update some tests to pass on desktop platforms. r=tn (afa54f4dc9)
     - Bug 1169666 - Revert reftest sanity flag ordering, fixes failures on OS X. (553743b4ce)
     - Bug 1114526. Add reftest. (ef2589e3b8)
     - Bug 1192616 - Skip over some reftests which fail on the pandaboards with the new dynamic toolbar implementation due to the screen size being too small. r=gbrown (fd3a0a523c)
     - Bug 1185747 part 2 - Remove magical reftest harness properties and use standard meta-viewport tags instead. r=tn (73d2d442d9)
     - Bug 1194811 Part 1 - Recompute the zoom constraints if the available screen area changes. r=botond (620dc82022)
     - Bug 1194811 Part 2 - Use the content viewer size rather than the composition size of the root frame when computing the CSS viewport. r=botond (089459fcb5)
     - Bug 1185747 part 3 - Rip out code to explicitly override the CSS viewport. r=tn (00ea1c7277)
     - Bug 1178354 - Ensure we fire a before-first-paint event for printing as well. r=tn (3dfd7f0f76)
     - Bug 1152254 - Handle vertical text frames when clipping display list for drag image. r=smontagu (965256a547)
     - Bug 1156135. Add runtime testing of graphics features. r=mattwoodrow,mossop (6a8cb24421)
     - Refactor the graphics sanity test to support multiple snapshots. (bug 1173117 part 1, r=mattwoodrow) (8a0a78e4d3)
     - Add an observer service notification for the first widget paint message. (bug 1173117 part 2, r=roc) (e421003dcd)
     - Fix a widget size check bug in nsWindow::CaptureWidgetOnScreen. (bug 1173117 part 3, r=mattwoodrow) (485694c380)
     - Add OS snapshotting to the gfx sanity test and report whether or not it matches the compositing test results. (bug 1173117 part 4, r=mattwoodrow,vladan) (38e82d10ad)
     - Bug 1191608 - initialize element to null in CanvasRenderingContext2D::DrawImage. r=bas (e26dd8b8ce) (e0e5f031a0)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1196041 - Disallow getter/setter with expression closure in class declaration. r=efaust (ee47aae93d)
     - Bug 1206485 - "Boot loop after first boot on some devices (Xperia M2, ...)" [r=terrence f=lissyx+mozillians] (3c73ad18a9)
     - Bug 1204368 - Fix modifier used for ASI after do-while. r=Waldo (857712ea07)
     - Bug 1189872 - Make {Map, Set}.prototype an ordinary object. r=Waldo (69abffd59b)
     - Bug 1199175 - Fix Debugger::slowPathOnLeaveFrame to remove the frame on OOM too. r=shu (35d190cac7)
     - code style (debd143914)
     - Bug 1133196 - Ensure script observability when setting Debugger.Frame.onStep. (r=jandem) (ce207a4762) (9c5061e940)
    - ported from `custom` branch of UXP: libstagefright: relax ctts flags checking. (d9f11187) (64bef5bfd2)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1204404 - Odin: move assert to avoid assertion failure. r=lth (14bd78e697)
     - Bug 1204864 - Odin: reject UINT32_MAX heap resize mask. r=bbouvier (7242554d2a)
     - Bug 1204847: Reinterpret the asmFunc pointer as an AsmFunction in case of offthread compile error; r=luke (b82f210a53) (0bc2879198)

    * Notice: the changelog above may not always applicable to XULRunner code which K-Meleon uses.

    A goanna3 source tree that has kmeleon adaption patch applied is available here: https://github.com/roytam1/palemoon27/tree/kmeleon76

    • Upvote 1

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New NewMoon 27 Build!

    32bit https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20220528-0bc2879198-xpmod.7z
    32bit SSE https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20220528-0bc2879198-xpmod-sse.7z
    32bit noSSE https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20220528-0bc2879198-xpmod-ia32.7z

    64bit https://o.rthost.win/palemoon/palemoon-27.10.0.win64-git-20220528-0bc2879198-xpmod.7z

    source repo: https://github.com/roytam1/palemoon27

    repo changes since my last build:
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1134252 - Don't crash the content process if RenderFrameParent is not constructed successfully. r=billm. (2564cb0e6a)
     - Bug 1180644: Fix crashes after enabling OOP on B2GDroid. r=snorp (d585c571e3)
     - Bug 1198674 - null-check mFrameLoader in RenderFrameParent. r=sotaro (86f26b2046)
     - Bug 1198674 - Null-check mFrameLoader before calling GetFrom in RenderFrameParent. r=sotaro (33bd495e75)
     - Bug 1200778 - Make sure to update the APZCTreeManager associated with a RenderFrameParent when it is dragged to a new window. r=mstange (bf2d25616c)
     - Bug 1202703 - Part 1 - CreateRenderingContext can fail. r=mattwoodrow (50de4cd050)
     - Bug 1185747 part 1 - Use pref/meta-viewport tag instead of DOMWindowUtils to set the CSS viewport for mochitests. r=tn (62a8c1d460)
     - Bug 1147038 - Update some tests to pass on desktop platforms. r=tn (afa54f4dc9)
     - Bug 1169666 - Revert reftest sanity flag ordering, fixes failures on OS X. (553743b4ce)
     - Bug 1114526. Add reftest. (ef2589e3b8)
     - Bug 1192616 - Skip over some reftests which fail on the pandaboards with the new dynamic toolbar implementation due to the screen size being too small. r=gbrown (fd3a0a523c)
     - Bug 1185747 part 2 - Remove magical reftest harness properties and use standard meta-viewport tags instead. r=tn (73d2d442d9)
     - Bug 1194811 Part 1 - Recompute the zoom constraints if the available screen area changes. r=botond (620dc82022)
     - Bug 1194811 Part 2 - Use the content viewer size rather than the composition size of the root frame when computing the CSS viewport. r=botond (089459fcb5)
     - Bug 1185747 part 3 - Rip out code to explicitly override the CSS viewport. r=tn (00ea1c7277)
     - Bug 1178354 - Ensure we fire a before-first-paint event for printing as well. r=tn (3dfd7f0f76)
     - Bug 1152254 - Handle vertical text frames when clipping display list for drag image. r=smontagu (965256a547)
     - Bug 1156135. Add runtime testing of graphics features. r=mattwoodrow,mossop (6a8cb24421)
     - Refactor the graphics sanity test to support multiple snapshots. (bug 1173117 part 1, r=mattwoodrow) (8a0a78e4d3)
     - Add an observer service notification for the first widget paint message. (bug 1173117 part 2, r=roc) (e421003dcd)
     - Fix a widget size check bug in nsWindow::CaptureWidgetOnScreen. (bug 1173117 part 3, r=mattwoodrow) (485694c380)
     - Add OS snapshotting to the gfx sanity test and report whether or not it matches the compositing test results. (bug 1173117 part 4, r=mattwoodrow,vladan) (38e82d10ad)
     - Bug 1191608 - initialize element to null in CanvasRenderingContext2D::DrawImage. r=bas (e26dd8b8ce) (e0e5f031a0)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1196041 - Disallow getter/setter with expression closure in class declaration. r=efaust (ee47aae93d)
     - Bug 1206485 - "Boot loop after first boot on some devices (Xperia M2, ...)" [r=terrence f=lissyx+mozillians] (3c73ad18a9)
     - Bug 1204368 - Fix modifier used for ASI after do-while. r=Waldo (857712ea07)
     - Bug 1189872 - Make {Map, Set}.prototype an ordinary object. r=Waldo (69abffd59b)
     - Bug 1199175 - Fix Debugger::slowPathOnLeaveFrame to remove the frame on OOM too. r=shu (35d190cac7)
     - code style (debd143914)
     - Bug 1133196 - Ensure script observability when setting Debugger.Frame.onStep. (r=jandem) (ce207a4762) (9c5061e940)
    - ported from `custom` branch of UXP: libstagefright: relax ctts flags checking. (d9f11187) (64bef5bfd2)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1204404 - Odin: move assert to avoid assertion failure. r=lth (14bd78e697)
     - Bug 1204864 - Odin: reject UINT32_MAX heap resize mask. r=bbouvier (7242554d2a)
     - Bug 1204847: Reinterpret the asmFunc pointer as an AsmFunction in case of offthread compile error; r=luke (b82f210a53) (0bc2879198)

    • Like 2

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New build of post-deprecated Serpent/moebius for XP!
    * Notice: This repo will not be built on regular schedule, and changes are experimental as usual.
    ** Current moebius patch level should be on par with 52.9, but some security patches can not be applied/ported due to source milestone differences between versions.

    Test binary:
    Win32 http://o.rthost.win/basilisk/basilisk55-win32-git-20220528-c952169c0-xpmod.7z
    Win64 http://o.rthost.win/basilisk/basilisk55-win64-git-20220528-c952169c0-xpmod.7z

    repo: https://github.com/roytam1/basilisk55

    Repo changes:
    - import from UXP: Issue #1899 - Disable the (broken) MDN integration widget by default. (a9d3c2b5) (badef41ee)
    - import from UXP: Issue #1899 - Make sure the test for it still works (5338dcd1) (dbdfa5c5a)
    - import from UXP: Issue #1813 - Enable date and time picker by default. (86db73b5) (897709ad7)
    - import from UXP: Issue #1210 - Keep timepicker disabled for now. (1bf41a59) (b2e9a3ed6)
    - ported from UXP: Bug 1679987 - Remove unused includes of nsCharSeparatedTokenizer.h. (62a140ab) (adb65da82)
    - import from UXP: Issue #1894 - Part 1: Implement coalesce JS opcode (3efa2347) (724f5386d)
    - import from UXP: Issue #1894 - Part 2: Implement support for nullish coalescing in the JS parser (d7cdeaf3) (3f52d9a1e)
    - import from UXP: Issue #1894 - Part 3: Implement support for nullish coalescing in JS reflection (f282a4c0) (9d96714a9)
    - ported from UXP: Issue #1894 - Part 4: Implement IonMonkey support for nullish coalescing (58127218) (e22fd60eb)
    - import from UXP: Issue #1894 - Part 5: Implement bytecode for nullish coalescing (707867d1) (ed1e4b60e)
    - import from UXP: Issue #1894 - Part 6: Check for nullish values when folding coalesce nodes (9dde32ee) (4c25f1d2e)
    - import from UXP: Issue #1894 - Part 7: Update tests (f0e208a8) (c953ce433)
    - import from UXP: No issue - Add null check to send packet function in the developer tools server (2e975416) (3f8cd2184)
    - ported from UXP: Issue #457 - Silence some GCC compiler warnings in FFmpeg code (5c61382d) (a9ce580ca)
    - import from `custom` branch of UXP: libstagefright: relax ctts flags checking. (d9f11187) (c952169c0)

    • Upvote 1

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New build of BOC/UXP for XP!

    Test binary:
    MailNews Win32 https://o.rthost.win/boc-uxp/mailnews.win32-20220528-3dbfc66a-uxp-0855ba43d-xpmod.7z
    BNavigator Win32 https://o.rthost.win/boc-uxp/bnavigator.win32-20220528-3dbfc66a-uxp-0855ba43d-xpmod.7z

    source repo (excluding UXP): https://github.com/roytam1/boc-uxp/tree/custom

    * Notice: the profile prefix (i.e. parent folder names) are also changed since 2020-08-15 build, you may rename their names before using new binaries when updating from builds before 2020-08-15.

    --

    New build of HBL-UXP for XP!

    Test binary:
    IceDove-UXP(mail) https://o.rthost.win/hbl-uxp/icedove.win32-20220528-id-181ea5d-uxp-0855ba43d-xpmod.7z
    IceApe-UXP(suite) https://o.rthost.win/hbl-uxp/iceape.win32-20220528-id-181ea5d-ia-c642e3c-uxp-0855ba43d-xpmod.7z

    source repo (excluding UXP):
    https://github.com/roytam1/icedove-uxp/tree/winbuild
    https://github.com/roytam1/iceape-uxp/tree/winbuild

    for UXP changes please see above.

    • Upvote 1

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New build of Serpent/UXP for XP!

    Test binary:
    Win32 https://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20220528-3219d2d-uxp-0855ba43d-xpmod.7z
    Win64 https://o.rthost.win/basilisk/basilisk52-g4.8.win64-git-20220528-3219d2d-uxp-0855ba43d-xpmod.7z

    source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/custom

    IA32 Win32 https://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20220528-3219d2d-uxp-0855ba43d-xpmod-ia32.7z

    source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/ia32

    NM28XP build:
    Win32 https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20220528-d849524bd-uxp-0855ba43d-xpmod.7z
    Win32 SSE https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20220528-d849524bd-uxp-0855ba43d-xpmod-sse.7z
    Win64 https://o.rthost.win/palemoon/palemoon-28.10.6a1.win64-git-20220528-d849524bd-uxp-0855ba43d-xpmod.7z

    Official UXP changes picked since my last build:
    - Bug 1679987 - Remove unused includes of nsCharSeparatedTokenizer.h. (62a140ab8)
    - Issue #1894 - Part 1: Implement coalesce JS opcode (3efa23472)
    - Issue #1894 - Part 2: Implement support for nullish coalescing in the JS parser (d7cdeaf31)
    - Issue #1894 - Part 3: Implement support for nullish coalescing in JS reflection (f282a4c05)
    - Issue #1894 - Part 4: Implement IonMonkey support for nullish coalescing (581272180)
    - Issue #1894 - Part 5: Implement bytecode for nullish coalescing (707867d16)
    - Issue #1894 - Part 6: Check for nullish values when folding coalesce nodes (9dde32ee9)
    - Issue #1894 - Part 7: Update tests (f0e208a86)
    - No issue - Add null check to send packet function in the developer tools server (2e975416b)
    - Issue #457 - Silence some GCC compiler warnings in FFmpeg code (5c61382da)
    - No issue - relax ctts flag checking in media/libstagefright (2f8131275)

    No official Pale-Moon changes picked since my last build.

    No official Basilisk changes picked since my last build.

    My changes picked since my last build:
    - libstagefright: relax ctts flags checking. (d9f111872)
    - [Basilisk] pdfjs: remove telemetry (0855ba43d)

    * Notice: From now on, UXP rev will point to `custom` branch of my UXP repo instead of MCP UXP repo, while "official UXP changes" shows only `tracking` branch changes.

    • Like 4

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted · Edited by roytam1

    3 hours ago, VistaLover said:

    libstagefright fix by @roytam1, committed more than two days ago, as a response to a bug report by yours truly ;) :

    libstagefright: relax ctts flags checking.

    libstagefright fix by Moonchild ("official" UXP), committed 21 hours ago:

    No issue - relax ctts flag checking in media/libstagefright

    MCP's code is almost identical...

    Conclusion/Question: Is Moonchild now following "roytam1/UXP" ? :P :whistle:

    because someone else reported in their forum: https://forum.palemoon.org/viewtopic.php?f=70&t=28405

    and for obvious reason, he rewrote the commit in his own words (maybe based on my comment but I don't actually know)

     

    if the reporter is viewing this thread, I would like to say: please tell MCP to support HEVC, Opus, and FLAC in MPEG4Extractor in libstagefright as well!

    36 minutes ago, Elkern 4926 said:

    Also, looks like M.A.T is keep pace: https://code.binaryoutcast.com/projects/aura-central/commits/TRUNK

    Will updates from him be added to BNavigator and others?

    if applicable, then yes.

    • Upvote 1

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    1 hour ago, UCyborg said:

    I disagree, that "solution" is like not fixing broken Bugatti Chiron and taking VW Bug for a drive instead. Sure you got to destination, but your Bugatti is still broken. Same with websites, you don't have "old" versions of many websites available. Also, someone else already wrote a similar user script - https://greasyfork.org/en/scripts/40897-old-reddit-please.

    UXP browsers will need to be updated to understand regex named capture groups. I doubt roytam1 is the correct address for such issues. I tried rewriting that part Reddit code, but got the code wrong (with help of palefill extension)...may take a look at it again this weekend. The script executes without syntax errors only when the site doesn't intend to hide most of discussion and show the button to view it.

    for regex engine you should watch https://repo.palemoon.org/MoonchildProductions/UXP/issues/1675

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted · Edited by roytam1

    26 minutes ago, NotHereToPlayGames said:

    I once knew a guy that ran a "Tor Network Server".  I cannot claim to know "how" that works, but I do know the guy running that server would tell story after story of how many people were downloading illegal software and movies through his "server".

    He swore up and down that he could not be held liable for the illegal actions occurring on his "server".  Again, cannot claim to know "how" that works.

    But I do suspect very highly that the people on this "Tor Network" had no clue that the owners of these "servers" had their IP Address and knew what they were downloading.

    better not to become a tor "exit". I did that in the past and police reached me (thay said that they found my IP uploaded child porn) and checked all my networked machines in an early morning. Several hours later they found nothing and told me not to become an exit or they will come again.

    • Upvote 1

    Browsing the web on 98/ME in 2019 and beyond

    in Windows 9x/ME

    Posted

    10 minutes ago, Dave-H said:

    @roytam1

    Hi Roy. This query is actually Windows 98 related, so perhaps shouldn't be here, but I didn't think it was worth starting another thread in the Windows 98 section.

    Some time ago, IIRC you created experimentally a couple of builds of old browsers which would work on Windows 98SE with updated security protocols.
    I've been using your builds of Firefox 3.6 Community Edition and K-Meleon 74 for quite some time now on Windows 98SE, and they have both worked fine, although there are some stability issues with K-Meleon.

    The only thing I regularly use Firefox 3.6 for is downloading weekly virus pattern updates from Trend Micro to keep my ancient installation of PC-cillin 2002 up to date.
    This has always worked fine until recently, but now although I can access the download page fine, when i try to actually download the file, I'm getting this -

    Image1.thumb.jpg.2d4a443e56e07bb6fff646ef25b1698a.jpg

    I have tried with K-Meleon, but that just freezes on the download page.
    However, if I copy the location of the pattern file from Firefox 3.6, and paste that into K-Meleon, I can then download the file!
    A workaround, but a bit awkward.
    Can anything be now done about the security error in Firefox 3.6, or is it now a lost cause?
    Strange that I can still access the download page itself, but not the file I need to download.
    Cheers, Dave.
    :)

    my 3.6 build is somewhat unmaintained and its security cypher suites may not suit what websites want:

    https://www.ssllabs.com/ssltest/analyze.html?d=www.trendmicro.com#hidecipher1

    but it seems that my fx3.6 build may be able to connect this site.

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    7 hours ago, VistaLover said:

    The 1s, video-only, clip (GIFV) in the tweet below:

    https://twitter.com/KeviSkillz/status/1527748243001581571

    refuses to play in Serpent 52.9.0 32-bit:

    srTWbg8.jpg

    I tried first with Vista SP2's native h264 decoder,

    media.wmf.enabled;true
    media.ffvpx.enabled;false

    then with St52's "embedded" h264 decoder,

    media.wmf.enabled;false
    media.ffvpx.enabled;true

    and then with BOTH turned on, but still no dice... :( The direct link to the clip is:

    https://video.twimg.com/tweet_video/FTOm21xX0AcJkcs.mp4

    and when that is loaded in a tab, it too, as you'd expect, fails to play, citing: "Video format or MIME type is not supported";
    perhaps UXP's MP4 parser and/or the (ffmpeg-based) h264 decoder of St52 (inside the ffvpx lib) is in need of an update/fix... :whistle:
    "naturally", 360EE has no problem playing back that embedded clip... :dubbio:

    the problem is in old libstagefright bundled in UXP as it can't find any tracks from that MP4 file. I need to dig upstreams to see if there is a fix for this issue.

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted · Edited by roytam1

    29 minutes ago, Egorkaru said:

    roytam1, can you still update RetroZilla Browser for Windows 9x based on Mozilla Firefox 2.0.0.20 and add support for the following cipher suites?:

    • TLS 1.2: ECDHE_ECDSA_AES_256_GCM_SHA384
    • TLS 1.3: AES_128_GCM_SHA256
    • TLS 1.3: AES_256_GCM_SHA384
    • TLS 1.3: CHACHA20_POLY1305_SHA256

    Then RetroZilla Browser would be the immortal browser for Windows 9x and Pentium 1.

    not possible without rewriting upstream changes in C89 syntax, thus not possible to be included in NSS-Legacy.

    my old answer still applies:

     

    • Upvote 2

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New regular/weekly KM-Goanna release:
    https://o.rthost.win/kmeleon/KM76.4.6-Goanna-20220521.7z

    Changelog:

    Out-of-tree changes:
    * update Goanna3 to git 646c78a9d0...e7535dc94b:
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1194059 (Part 1) - Ensure that metadata decode progress is always delivered atomically. r=tn (07f0441600)
     - Bug 1191090 - Use the normal PNG decoder for PNG metadata decodes. r=tn (ce3fe1be5f)
     - Bug 1191114 (Part 1) - Always detect HAS_TRANSPARENCY during the metadata decode. r=tn (3841132932)
     - Bug 1191114 (Part 2) - Add support for creating an anonymous metadata decoder, for use in tests. r=tn (2cdcc0c278)
     - Bug 1191114 (Part 3) - Add flags to image test cases. r=tn (4a6f5a5230)
     - Bug 1191114 (Part 4) - Add tests for metadata decoding, including that we always deliver HAS_TRANSPARENCY during the metadata decode. r=tn (b9c5d1cd4a)
     - Bug 1126330 - Remove the check for non-looping animations. r=seth (828dabba24)
     - Bug 1194059 (Part 2) - Always detect IS_ANIMATED during the metadatadecode. r=tn (0ba5bf38f1)
     - Bug 1194059 (Part 3) - Ensure the nsIInputStream LoadImage() returns is always buffered. r=tn (ed2b02205b)
     - Bug 1194059 (Part 4) - Add tests that we detect IS_ANIMATED during the metadata decode. r=tn (298f14a7c9)
     - Bug 1188705 (part 1) - Remove gfxASurface::GetMemoryLocation(). r=seth. (1f0da73a08)
     - Bug 1188705 (part 2) - Remove unused SizeOfDecodedWithComputedFallbackIfHeap declaration. r=seth. (3356dbed06)
     - Bug 1188705 (part 3) - Simplify imgFrame::SizeOfExcludingThis(). r=seth. (563262a834)
     - Bug 1155252 - Don't allocate X11TextureClients bigger than xlib's maximum surface size. r=jrmuizel (3f11590667)
     - Bug 1143994 - Fix some -Wunreachable-code and -Wswitch warnings in imagelib. r=seth (008becc7e2)
     - Bug 1060609 (Part 1) - Disable downscale-during-decode when HQ scaling is disabled. r=tn (6da77e3cad)
     - Bug 1187569 - PNGs getting stuck in a pixelated state. r=seth (da305ef99c)
     - Bug 1194900 - Stop deciding when to send invalidations in nsPNGDecoder and let Decoder handle it. r=tn (50fa14a984)
     - Bug 1151694 - Part 1 - Move CommonAnimationManager::sLayerAnimationInfo into LayerAnimationInfo.(cpp|h). r=bbirtles (9f93e0d569)
     - Bug 1151694 - Part 2 - imgTools should be inside mozilla::image namespace. r=bbirtles (8dfc3f2e4b)
     - Bug 1196066 (Part 1) - Fix bad directory entries in two of our ICO reftests. r=tn (9e4c70d2b4)
     - Bug 1196065 - Add sanity tests for image decoders. r=tn (557b9131cb)
     - Bug 1194912 (Part 1) - Add CopyOnWrite<T> to support automatic copy-on-write for recursive writes to data structures. r=tn (b081a50716)
     - Bug 1196066 (Part 2) - Add a streaming lexing framework to ImageLib. r=tn (59eb634ea5)
     - Bug 1196476 - Replace ProgressTracker::FirstObserverIs() with a simpler mechanism on imgRequest. r=tn (db9ecc65ef)
     - missing part of Bug 1139225 (Part 2) - Dispatch OnImageAvailable to the main thread manually in imgRequest. r=tn (e7b22db614)
     - Bug 1194912 (Part 2) - Store ProgressTracker observers in a copy-on-write hash table, and dispatch notifications to them using a template. r=tn (5efd7b38b3)
     - Bug 1180225. Make convolver more like upstream. r=seth (18e3c168fc)
     - Bug 1149318 - Fix the calling convention on SkGetUserDefaultLocaleNameProc. r=eihrul (7b750d4e4e)
     - Bug 1210493 - enlarge stroke bounds by line width when doing a quick-reject in SkDraw::drawRect. r=jmuizelaar (e8b5d0fe2d)
     - Bug 1188206 - Fix more constructors in gfx; r=jrmuizel (944ea9938c) (b75b514a74)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - namespace (dd7e1c593d)
     - Bug 1158541 - Remove TiledTextureImage::mSize; r=jrmuizel This member is never initialized to anything so it always contains a (0,0) size, and it shadows the protected mSize member in the base class which is actually initialized to the correct size. (e2930b05a8)
     - Bug 1158542 - Remove TextureImage::mImageFormat; r=jrmuizel This is never initialized or accessed. (c5af3763b5)
     - Bug 1136428 - Create a different set and restore path of the draw buffer state in WebGLContext::ForceClearFramebufferWithDefaultValues depending on whether or not the default framebuffer is being used. r=jgilbert (e92fbcaa38)
     - Bug 1136428 - Change implementation of WebGLContext::DrawBuffers such that it goes through GLScreenBuffer to change the draw buffer. Added a variable to GLScreenBuffer to cache the draw buffer mode. r=jgilbert (eb80c3ac18)
     - Bug 1136428 - Ensure we never create more than the amount supported of color attachments in WebGLFrameBuffer or try to get an attachment point higher than the supported number. r=jgilbert (3050ce6979)
     - Bug 1185803 - Replace MOZ_CRASH with GenerateWarning. r=jgilbert (1b5ba983c6)
     - Bug 1188540 - Forward texture-related functions to WebGLTexture. - r=kamidphish (10f84c83e5)
     - Bug 1189903 - Don't use RGBA surfaces on GLX if surface sharing is not used. r=jgilbert (2edbca3d13)
     - Bug 1170842 - Part 3: Implement GetInternalformatParameter. r=jgilbert, r=smaug (18f0fdf5f9)
     - Bug 1170842 - Part 4: Implement FramebufferTextureLayer. r=jgilbert (2acd7963e7)
     - Bug 1062066 (Part 1) - Add support for vertically flipping downscaler output. r=tn (f23315cba2)
     - Bug 1204626 - Add a regression test. r=billm (aa3c694b83)
     - Bug 1062066 (Part 2) - Add a Downscaler API to clear a row. r=tn (7c77bf27b0)
     - Bug 1062066 (Part 3) - Only mark BMP surfaces as transparent if they actually have alpha data. r=tn (ed9bade04e)
     - Bug 1060609 (Part 2) - Add downscale-during-decode support for the PNG decoder. r=tn,f=glennrp (47259a797a)
     - hack to fix build (12266daf48)
     - Bug 1060609 (Part 3) - Add tests that interlaced and non-interlaced PNGs have the same downscaling behavior. r=tn (763b18b15e)
     - Bug 1062066 (Part 4) - Add downscale-during-decode support for the BMP decoder. r=tn (5a810ed168)
     - indentation fixes (872c66d3a8) (00b0a024a4)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1188569: Drop unneeded MOZ_WARN_UNUSED_RESULT from from LookupBestMatch in SurfaceCache.cpp. r=seth (5e74e0028c)
     - Bug 1192356 (Part 1) - Take advantage of mozilla::Tie() in SurfaceCache.cpp. r=dholbert (e4908c725d)
     - Bug 1192356 (Part 2) - Take advantage of mozilla::Tie() in RasterImage.cpp. r=tn (1204189b73)
     - Bug 1185800 - Add DecoderFlags and SurfaceFlags enum classes and use them instead of imgIContainer flags in all decoder-related code. r=tn (3abdab11f6)
     - Bug 1196066 (Part 3) - Rewrite nsICODecoder to use StreamingLexer. r=tn (e2ba590c9d)
     - Bug 1196066 (Part 4) - Enable the ICOMultiChunk test, which now passes. r=tn (9e02611959)
     - Bug 1124084 - Flip on downscale-during-decode everywhere. r=tn (bd9deff966)
     - Bug 1160801 - Treat invalid GIF disposal methods as DisposalMethod::NOT_SPECIFIED. r=jrmuizel (e26feaf8fb)
     - Bug 1201796 (Part 1) - Treat ICOs with wrong widths and heights as corrupt. r=tn (322ba20808)
     - Bug 1201796 (Part 2) - Add GetFrameAtSize() to support downscale-during-decode for GetFrame() use cases. r=tn (92f5d3a0a7)
     - Bug 1194906 - Replace 'NS_ENSURE_TRUE(BadImage(..))' warnings with more useful messages. r=tn (cc3b368673)
     - Bug 1201796 (Part 3) - Enable downscale-during-decode for imgITools::EncodeScaledImage(). r=tn (e2cdb5b520)
     - Bug 1194472 - Correctly fetch compositor backend in WebGLContext. r=jgilbert (0092052dfc)
     - Bug 1161913 - Part 1 - Add invalidation state for CaptureStream to Canvas and Contexts. r=mt (0377d6bbe7)
     - Bug 1168075 - Fix CanvasCaptureMediaStream build fail for bluetooth2. r=pehrsons (53c67c0056)
     - Bug 1176363 - Part 1: Make a raw copy of each Canvas::CaptureStream frame. r=mattwoodrow (a5df5793d6)
     - Bug 1194575 - Rename RecoverFromLossOfFrames() to RecoverFromInvalidFrames() to better reflect its role. r=tn (baa6455e79)
     - Bug 1146663 (Part 1) - Remove HQ scaling, which is now dead code. r=tn (efaddadea0)
     - Bug 1146663 (Part 2) - Remove the concept of lifetimes from the SurfaceCache. r=dholbert (ab9862d7ee)
     - Bug 1146663 (Part 3) - Make it impossible to deoptimize imgFrames. r=tn (19e2f1b370)
     - Bug 1201763 - Add downscale-during-decode support for the ICON decoder. r=tn (33a2b95e5c)
     - Bug 1194058 (Part 1) - Add Deinterlacer to allow Downscaler to work with interlaced images. r=tn (f7c57b7a8e)
     - Bug 1194058 (Part 2) - Add downscale-during-decode support for the GIF decoder. r=tn (85622f9d55)
     - Bug 1201796 (Part 4) - Add downscale-during-decode support for the ICO decoder. r=tn (d09d18b0d9)
     - Bug 1146663 (Part 4) - Make all RasterImages support downscale-during-decode. r=tn (264642a895)
     - Bug 1146663 (Part 5) - Require that all image decoders support downscale-during-decode. r=tn (79ad99885d)
     - Bug 1206836 - When downscaling ICOs, downscale the AND mask as well. r=tn a=KWierso (08ec3d092b)
     - missing bit of Bug 1138293 - Use malloc/free/realloc/calloc (eb8e5e1b9c)
     - missing bit of Bug 1146663 (Part 3) - Make it impossible to deoptimize imgFrames. (233befe48f)
     - Bug 1208935 - Move Deinterlacer to a standalone file. r=seth (b50322abc286) (f1d1e16669)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1187401 (Part 1) - Simplify the condition that determines whether we set RasterImage::mHasBeenDecoded. r=tn (69be36e7ef)
     - Bug 1187401 (Part 2) - Eliminate the nsresult return value from RasterImage::SetMetadata, since it's not used anymore. r=tn (dc521c4b9f)
     - Bug 1187401 (Part 3) - For consistency, call DoError if SetMetadata sees a negative size. r=tn (d9ca8ec91b)
     - Bug 1207183 - micro-optimize removing work items from DecodePool's queues; r=seth (6bd2717e3a)
     - No bug - Fix out-of-date comment in Decoder.cpp. r=me (54fdbbd444)
     - Bug 1181324 - Eliminate the duplicate mRefCnt member in MultipartImage. r=seth (c4512a443e)
     - Bug 1180105 - Do not leak the SourceSurface returned from imgIContainer::GetFrame in BlockUntilDecodedAndFinishObserving; r=seth (1cadbffc53)
     - Bug 1181909 - Fix potential null dereference in NextPartObserver. r=tn (cfd8ad0119)
     - Bug 108603 - Remove NS_IMPL_QUERY_INTERFACE_INHERITED0. r=mccr8, r=froydnj (4bfa6771fc)
     - Bug 1159502 - Don't block onload for multipart images. r=tn (8b50eadf39)
     - Bug 1200413 - Part 1: Re-write RefCountedInsideLambdaChecker to use captures instead of checking for DeclRef instances, r=ehsan (80ef99efe2)
     - Bug 1200413 - Part 2: Make lambdas in ProgressTracker.cpp capture strong references, r=seth (9e4d96dffa)
     - Bug 1194557 - Ensure that if the image was locked before RecoverFromLossOfFrames() was called, it's still locked afterwards. r=tn (ea4dc6ea9f)
     - code style (ad3773ba42)
     - Bug 1167590 - Mark imgRequestProxy::mListener as MOZ_UNSAFE_REF. r=seth (946ffaed8a)
     - Bug 1148397 - Fix data race on imgRequest::mHadInsecureRedirect. r=tanvi (e73d0664f3)
     - No bug - Tweak formatting of logging statement in imgRequest. r=me (cab2bcb014)
     - Bug 1180126 - Read content disposition regardless of content type in imgRequest::PrepareForNewPart. r=tn (2934597743)
     - Bug 1139225 - Followup - Remove duplicate multiPartChannel variable. (7f7f555a0b)
     - Bug 1141398 - Do not always revalidate image cache entries for file URIs. r=tn (31d73cb508)
     - Bug 1183563 - Fix incorrect mixed content warning after internal redirects. r=tanvi, r=seth (12a6c8a15f)
     - Bug 1150127 - Stop leaking windows via imgCacheValidator. r=baku (a7809c5fa7)
     - bits of Bug 1102048 (Part 20, imgLoader) (b2098c8a5a)
     - (No bug) - Correct blatantly lying comment in imgLoader.cpp. r=me DONTBUILD (64c42a5b09)
     - Bug 1160592 - Report image source size again in about:memory. r=dholbert (4e04cf3c3e)
     - Add an assertion for the first argument of NewImageChannel, no bug (4c8f087a8f)
     - Bug 1127534 - Remove assertion before creating a channel (r=sicking) (988692dc91)
     - Bug 1175371 - Make VectorImage wait to deliver LOAD_COMPLETE until its size is available. r=dholbert (3c81e0daff)
     - Bug 1181323 - Move nsSVGRenderingObserver's isupports/refcounting decl to subclasses, since one subclass (nsSVGFilterReference) already has its own redundant copy of the decl. r=dholbert (6171171c2c)
     - Bug 1161722 - If we're shutting down, don't warn about untracked unlocked surfaces. r=dholbert (f7e18ce481)
     - Bug 1170877 - Track how many times the SurfaceCache has overflowed and report it in about:memory. r=dholbert (884176cb1d)
     - Bug 1161743 - Upgrade 'WARNING: Not expiration-tracking an unlocked surface' to an assertion. r=dholbert (9900169e7b)
     - Bug 1167557 - Crash when a null surface is passed to SurfaceCache::Insert. r=dholbert (b3c4cf60aa)
     - remove bypass cache not fonud either in FF nor TF (3ed4056a27)
     - Missing bit Bug 1102048 (Part 25, header guards)
     - Make image/src files comply (cb8ed2428f)
     - No bug - Remove obsolete comment in SourceBuffer.h. r=me (6e9c233448)
     - coding style (94b7269690)
     - Bug 1157065 - GFX: 2D: Add Loongson3 MMI helpers. r=jrmuizel (ebce946c91)
     - reverto to FF52 and TFF settings (e147a8c7b5)
     - Bug 1134599 - Fix rpi build target. r=jrmuizel, r=shuang (b9722f860c)
     - Bug 1129147 - Part 1. Take CanvasPath into a separate file, to avoid circular dependency. r=roc (859bcad807)
     - Bug 1129147 - Part 2. Path option to addHitRegion. r=ehsan r=gw280 (b2ab08a8a1)
     - Bug 1206076: Use a specialized PersistentBufferProvider for Canvas2D when using a SkiaGL DrawTarget. r=jrmuizel (859589caf8)
     - Bug 1188752 - Addendum: Make PersistentBufferProviderBasic constructor explicit. r=bustage on a CLOSED TREE (a27a4dc974)
     - style (72a65dcb26)
     - Bug 1198574 - Remove unnecessary argument for PersistentBufferProvider. r=bas (dca718bba8)
     - Bug 1163124 - The initial value of the canvas filter property should be "none". r=roc (59df6a01d8) (e7535dc94b)

    * Notice: the changelog above may not always applicable to XULRunner code which K-Meleon uses.

    A goanna3 source tree that has kmeleon adaption patch applied is available here: https://github.com/roytam1/palemoon27/tree/kmeleon76

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New NewMoon 27 Build!

    32bit https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20220521-e7535dc94b-xpmod.7z
    32bit SSE https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20220521-e7535dc94b-xpmod-sse.7z
    32bit noSSE https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20220521-e7535dc94b-xpmod-ia32.7z

    64bit https://o.rthost.win/palemoon/palemoon-27.10.0.win64-git-20220521-e7535dc94b-xpmod.7z

    source repo: https://github.com/roytam1/palemoon27

    repo changes since my last build:
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1194059 (Part 1) - Ensure that metadata decode progress is always delivered atomically. r=tn (07f0441600)
     - Bug 1191090 - Use the normal PNG decoder for PNG metadata decodes. r=tn (ce3fe1be5f)
     - Bug 1191114 (Part 1) - Always detect HAS_TRANSPARENCY during the metadata decode. r=tn (3841132932)
     - Bug 1191114 (Part 2) - Add support for creating an anonymous metadata decoder, for use in tests. r=tn (2cdcc0c278)
     - Bug 1191114 (Part 3) - Add flags to image test cases. r=tn (4a6f5a5230)
     - Bug 1191114 (Part 4) - Add tests for metadata decoding, including that we always deliver HAS_TRANSPARENCY during the metadata decode. r=tn (b9c5d1cd4a)
     - Bug 1126330 - Remove the check for non-looping animations. r=seth (828dabba24)
     - Bug 1194059 (Part 2) - Always detect IS_ANIMATED during the metadatadecode. r=tn (0ba5bf38f1)
     - Bug 1194059 (Part 3) - Ensure the nsIInputStream LoadImage() returns is always buffered. r=tn (ed2b02205b)
     - Bug 1194059 (Part 4) - Add tests that we detect IS_ANIMATED during the metadata decode. r=tn (298f14a7c9)
     - Bug 1188705 (part 1) - Remove gfxASurface::GetMemoryLocation(). r=seth. (1f0da73a08)
     - Bug 1188705 (part 2) - Remove unused SizeOfDecodedWithComputedFallbackIfHeap declaration. r=seth. (3356dbed06)
     - Bug 1188705 (part 3) - Simplify imgFrame::SizeOfExcludingThis(). r=seth. (563262a834)
     - Bug 1155252 - Don't allocate X11TextureClients bigger than xlib's maximum surface size. r=jrmuizel (3f11590667)
     - Bug 1143994 - Fix some -Wunreachable-code and -Wswitch warnings in imagelib. r=seth (008becc7e2)
     - Bug 1060609 (Part 1) - Disable downscale-during-decode when HQ scaling is disabled. r=tn (6da77e3cad)
     - Bug 1187569 - PNGs getting stuck in a pixelated state. r=seth (da305ef99c)
     - Bug 1194900 - Stop deciding when to send invalidations in nsPNGDecoder and let Decoder handle it. r=tn (50fa14a984)
     - Bug 1151694 - Part 1 - Move CommonAnimationManager::sLayerAnimationInfo into LayerAnimationInfo.(cpp|h). r=bbirtles (9f93e0d569)
     - Bug 1151694 - Part 2 - imgTools should be inside mozilla::image namespace. r=bbirtles (8dfc3f2e4b)
     - Bug 1196066 (Part 1) - Fix bad directory entries in two of our ICO reftests. r=tn (9e4c70d2b4)
     - Bug 1196065 - Add sanity tests for image decoders. r=tn (557b9131cb)
     - Bug 1194912 (Part 1) - Add CopyOnWrite<T> to support automatic copy-on-write for recursive writes to data structures. r=tn (b081a50716)
     - Bug 1196066 (Part 2) - Add a streaming lexing framework to ImageLib. r=tn (59eb634ea5)
     - Bug 1196476 - Replace ProgressTracker::FirstObserverIs() with a simpler mechanism on imgRequest. r=tn (db9ecc65ef)
     - missing part of Bug 1139225 (Part 2) - Dispatch OnImageAvailable to the main thread manually in imgRequest. r=tn (e7b22db614)
     - Bug 1194912 (Part 2) - Store ProgressTracker observers in a copy-on-write hash table, and dispatch notifications to them using a template. r=tn (5efd7b38b3)
     - Bug 1180225. Make convolver more like upstream. r=seth (18e3c168fc)
     - Bug 1149318 - Fix the calling convention on SkGetUserDefaultLocaleNameProc. r=eihrul (7b750d4e4e)
     - Bug 1210493 - enlarge stroke bounds by line width when doing a quick-reject in SkDraw::drawRect. r=jmuizelaar (e8b5d0fe2d)
     - Bug 1188206 - Fix more constructors in gfx; r=jrmuizel (944ea9938c) (b75b514a74)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - namespace (dd7e1c593d)
     - Bug 1158541 - Remove TiledTextureImage::mSize; r=jrmuizel This member is never initialized to anything so it always contains a (0,0) size, and it shadows the protected mSize member in the base class which is actually initialized to the correct size. (e2930b05a8)
     - Bug 1158542 - Remove TextureImage::mImageFormat; r=jrmuizel This is never initialized or accessed. (c5af3763b5)
     - Bug 1136428 - Create a different set and restore path of the draw buffer state in WebGLContext::ForceClearFramebufferWithDefaultValues depending on whether or not the default framebuffer is being used. r=jgilbert (e92fbcaa38)
     - Bug 1136428 - Change implementation of WebGLContext::DrawBuffers such that it goes through GLScreenBuffer to change the draw buffer. Added a variable to GLScreenBuffer to cache the draw buffer mode. r=jgilbert (eb80c3ac18)
     - Bug 1136428 - Ensure we never create more than the amount supported of color attachments in WebGLFrameBuffer or try to get an attachment point higher than the supported number. r=jgilbert (3050ce6979)
     - Bug 1185803 - Replace MOZ_CRASH with GenerateWarning. r=jgilbert (1b5ba983c6)
     - Bug 1188540 - Forward texture-related functions to WebGLTexture. - r=kamidphish (10f84c83e5)
     - Bug 1189903 - Don't use RGBA surfaces on GLX if surface sharing is not used. r=jgilbert (2edbca3d13)
     - Bug 1170842 - Part 3: Implement GetInternalformatParameter. r=jgilbert, r=smaug (18f0fdf5f9)
     - Bug 1170842 - Part 4: Implement FramebufferTextureLayer. r=jgilbert (2acd7963e7)
     - Bug 1062066 (Part 1) - Add support for vertically flipping downscaler output. r=tn (f23315cba2)
     - Bug 1204626 - Add a regression test. r=billm (aa3c694b83)
     - Bug 1062066 (Part 2) - Add a Downscaler API to clear a row. r=tn (7c77bf27b0)
     - Bug 1062066 (Part 3) - Only mark BMP surfaces as transparent if they actually have alpha data. r=tn (ed9bade04e)
     - Bug 1060609 (Part 2) - Add downscale-during-decode support for the PNG decoder. r=tn,f=glennrp (47259a797a)
     - hack to fix build (12266daf48)
     - Bug 1060609 (Part 3) - Add tests that interlaced and non-interlaced PNGs have the same downscaling behavior. r=tn (763b18b15e)
     - Bug 1062066 (Part 4) - Add downscale-during-decode support for the BMP decoder. r=tn (5a810ed168)
     - indentation fixes (872c66d3a8) (00b0a024a4)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1188569: Drop unneeded MOZ_WARN_UNUSED_RESULT from from LookupBestMatch in SurfaceCache.cpp. r=seth (5e74e0028c)
     - Bug 1192356 (Part 1) - Take advantage of mozilla::Tie() in SurfaceCache.cpp. r=dholbert (e4908c725d)
     - Bug 1192356 (Part 2) - Take advantage of mozilla::Tie() in RasterImage.cpp. r=tn (1204189b73)
     - Bug 1185800 - Add DecoderFlags and SurfaceFlags enum classes and use them instead of imgIContainer flags in all decoder-related code. r=tn (3abdab11f6)
     - Bug 1196066 (Part 3) - Rewrite nsICODecoder to use StreamingLexer. r=tn (e2ba590c9d)
     - Bug 1196066 (Part 4) - Enable the ICOMultiChunk test, which now passes. r=tn (9e02611959)
     - Bug 1124084 - Flip on downscale-during-decode everywhere. r=tn (bd9deff966)
     - Bug 1160801 - Treat invalid GIF disposal methods as DisposalMethod::NOT_SPECIFIED. r=jrmuizel (e26feaf8fb)
     - Bug 1201796 (Part 1) - Treat ICOs with wrong widths and heights as corrupt. r=tn (322ba20808)
     - Bug 1201796 (Part 2) - Add GetFrameAtSize() to support downscale-during-decode for GetFrame() use cases. r=tn (92f5d3a0a7)
     - Bug 1194906 - Replace 'NS_ENSURE_TRUE(BadImage(..))' warnings with more useful messages. r=tn (cc3b368673)
     - Bug 1201796 (Part 3) - Enable downscale-during-decode for imgITools::EncodeScaledImage(). r=tn (e2cdb5b520)
     - Bug 1194472 - Correctly fetch compositor backend in WebGLContext. r=jgilbert (0092052dfc)
     - Bug 1161913 - Part 1 - Add invalidation state for CaptureStream to Canvas and Contexts. r=mt (0377d6bbe7)
     - Bug 1168075 - Fix CanvasCaptureMediaStream build fail for bluetooth2. r=pehrsons (53c67c0056)
     - Bug 1176363 - Part 1: Make a raw copy of each Canvas::CaptureStream frame. r=mattwoodrow (a5df5793d6)
     - Bug 1194575 - Rename RecoverFromLossOfFrames() to RecoverFromInvalidFrames() to better reflect its role. r=tn (baa6455e79)
     - Bug 1146663 (Part 1) - Remove HQ scaling, which is now dead code. r=tn (efaddadea0)
     - Bug 1146663 (Part 2) - Remove the concept of lifetimes from the SurfaceCache. r=dholbert (ab9862d7ee)
     - Bug 1146663 (Part 3) - Make it impossible to deoptimize imgFrames. r=tn (19e2f1b370)
     - Bug 1201763 - Add downscale-during-decode support for the ICON decoder. r=tn (33a2b95e5c)
     - Bug 1194058 (Part 1) - Add Deinterlacer to allow Downscaler to work with interlaced images. r=tn (f7c57b7a8e)
     - Bug 1194058 (Part 2) - Add downscale-during-decode support for the GIF decoder. r=tn (85622f9d55)
     - Bug 1201796 (Part 4) - Add downscale-during-decode support for the ICO decoder. r=tn (d09d18b0d9)
     - Bug 1146663 (Part 4) - Make all RasterImages support downscale-during-decode. r=tn (264642a895)
     - Bug 1146663 (Part 5) - Require that all image decoders support downscale-during-decode. r=tn (79ad99885d)
     - Bug 1206836 - When downscaling ICOs, downscale the AND mask as well. r=tn a=KWierso (08ec3d092b)
     - missing bit of Bug 1138293 - Use malloc/free/realloc/calloc (eb8e5e1b9c)
     - missing bit of Bug 1146663 (Part 3) - Make it impossible to deoptimize imgFrames. (233befe48f)
     - Bug 1208935 - Move Deinterlacer to a standalone file. r=seth (b50322abc286) (f1d1e16669)
    - import changes from `dev' branch of rmottola/Arctic-Fox:
     - Bug 1187401 (Part 1) - Simplify the condition that determines whether we set RasterImage::mHasBeenDecoded. r=tn (69be36e7ef)
     - Bug 1187401 (Part 2) - Eliminate the nsresult return value from RasterImage::SetMetadata, since it's not used anymore. r=tn (dc521c4b9f)
     - Bug 1187401 (Part 3) - For consistency, call DoError if SetMetadata sees a negative size. r=tn (d9ca8ec91b)
     - Bug 1207183 - micro-optimize removing work items from DecodePool's queues; r=seth (6bd2717e3a)
     - No bug - Fix out-of-date comment in Decoder.cpp. r=me (54fdbbd444)
     - Bug 1181324 - Eliminate the duplicate mRefCnt member in MultipartImage. r=seth (c4512a443e)
     - Bug 1180105 - Do not leak the SourceSurface returned from imgIContainer::GetFrame in BlockUntilDecodedAndFinishObserving; r=seth (1cadbffc53)
     - Bug 1181909 - Fix potential null dereference in NextPartObserver. r=tn (cfd8ad0119)
     - Bug 108603 - Remove NS_IMPL_QUERY_INTERFACE_INHERITED0. r=mccr8, r=froydnj (4bfa6771fc)
     - Bug 1159502 - Don't block onload for multipart images. r=tn (8b50eadf39)
     - Bug 1200413 - Part 1: Re-write RefCountedInsideLambdaChecker to use captures instead of checking for DeclRef instances, r=ehsan (80ef99efe2)
     - Bug 1200413 - Part 2: Make lambdas in ProgressTracker.cpp capture strong references, r=seth (9e4d96dffa)
     - Bug 1194557 - Ensure that if the image was locked before RecoverFromLossOfFrames() was called, it's still locked afterwards. r=tn (ea4dc6ea9f)
     - code style (ad3773ba42)
     - Bug 1167590 - Mark imgRequestProxy::mListener as MOZ_UNSAFE_REF. r=seth (946ffaed8a)
     - Bug 1148397 - Fix data race on imgRequest::mHadInsecureRedirect. r=tanvi (e73d0664f3)
     - No bug - Tweak formatting of logging statement in imgRequest. r=me (cab2bcb014)
     - Bug 1180126 - Read content disposition regardless of content type in imgRequest::PrepareForNewPart. r=tn (2934597743)
     - Bug 1139225 - Followup - Remove duplicate multiPartChannel variable. (7f7f555a0b)
     - Bug 1141398 - Do not always revalidate image cache entries for file URIs. r=tn (31d73cb508)
     - Bug 1183563 - Fix incorrect mixed content warning after internal redirects. r=tanvi, r=seth (12a6c8a15f)
     - Bug 1150127 - Stop leaking windows via imgCacheValidator. r=baku (a7809c5fa7)
     - bits of Bug 1102048 (Part 20, imgLoader) (b2098c8a5a)
     - (No bug) - Correct blatantly lying comment in imgLoader.cpp. r=me DONTBUILD (64c42a5b09)
     - Bug 1160592 - Report image source size again in about:memory. r=dholbert (4e04cf3c3e)
     - Add an assertion for the first argument of NewImageChannel, no bug (4c8f087a8f)
     - Bug 1127534 - Remove assertion before creating a channel (r=sicking) (988692dc91)
     - Bug 1175371 - Make VectorImage wait to deliver LOAD_COMPLETE until its size is available. r=dholbert (3c81e0daff)
     - Bug 1181323 - Move nsSVGRenderingObserver's isupports/refcounting decl to subclasses, since one subclass (nsSVGFilterReference) already has its own redundant copy of the decl. r=dholbert (6171171c2c)
     - Bug 1161722 - If we're shutting down, don't warn about untracked unlocked surfaces. r=dholbert (f7e18ce481)
     - Bug 1170877 - Track how many times the SurfaceCache has overflowed and report it in about:memory. r=dholbert (884176cb1d)
     - Bug 1161743 - Upgrade 'WARNING: Not expiration-tracking an unlocked surface' to an assertion. r=dholbert (9900169e7b)
     - Bug 1167557 - Crash when a null surface is passed to SurfaceCache::Insert. r=dholbert (b3c4cf60aa)
     - remove bypass cache not fonud either in FF nor TF (3ed4056a27)
     - Missing bit Bug 1102048 (Part 25, header guards)
     - Make image/src files comply (cb8ed2428f)
     - No bug - Remove obsolete comment in SourceBuffer.h. r=me (6e9c233448)
     - coding style (94b7269690)
     - Bug 1157065 - GFX: 2D: Add Loongson3 MMI helpers. r=jrmuizel (ebce946c91)
     - reverto to FF52 and TFF settings (e147a8c7b5)
     - Bug 1134599 - Fix rpi build target. r=jrmuizel, r=shuang (b9722f860c)
     - Bug 1129147 - Part 1. Take CanvasPath into a separate file, to avoid circular dependency. r=roc (859bcad807)
     - Bug 1129147 - Part 2. Path option to addHitRegion. r=ehsan r=gw280 (b2ab08a8a1)
     - Bug 1206076: Use a specialized PersistentBufferProvider for Canvas2D when using a SkiaGL DrawTarget. r=jrmuizel (859589caf8)
     - Bug 1188752 - Addendum: Make PersistentBufferProviderBasic constructor explicit. r=bustage on a CLOSED TREE (a27a4dc974)
     - style (72a65dcb26)
     - Bug 1198574 - Remove unnecessary argument for PersistentBufferProvider. r=bas (dca718bba8)
     - Bug 1163124 - The initial value of the canvas filter property should be "none". r=roc (59df6a01d8) (e7535dc94b)

    • Like 2

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    New build of BOC/UXP for XP!

    Test binary:
    MailNews Win32 https://o.rthost.win/boc-uxp/mailnews.win32-20220521-3dbfc66a-uxp-1e871780f-xpmod.7z
    BNavigator Win32 https://o.rthost.win/boc-uxp/bnavigator.win32-20220521-3dbfc66a-uxp-1e871780f-xpmod.7z

    source repo (excluding UXP): https://github.com/roytam1/boc-uxp/tree/custom

    * Notice: the profile prefix (i.e. parent folder names) are also changed since 2020-08-15 build, you may rename their names before using new binaries when updating from builds before 2020-08-15.

    --

    New build of HBL-UXP for XP!

    Test binary:
    IceDove-UXP(mail) https://o.rthost.win/hbl-uxp/icedove.win32-20220521-id-181ea5d-uxp-1e871780f-xpmod.7z
    IceApe-UXP(suite) https://o.rthost.win/hbl-uxp/iceape.win32-20220521-id-181ea5d-ia-c642e3c-uxp-1e871780f-xpmod.7z

    source repo (excluding UXP):
    https://github.com/roytam1/icedove-uxp/tree/winbuild
    https://github.com/roytam1/iceape-uxp/tree/winbuild

    for UXP changes please see above.

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted · Edited by roytam1

    New build of Serpent/UXP for XP!

    Test binary:
    Win32 https://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20220521-3219d2d-uxp-1e871780f-xpmod.7z
    Win64 https://o.rthost.win/basilisk/basilisk52-g4.8.win64-git-20220521-3219d2d-uxp-1e871780f-xpmod.7z

    source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/custom

    IA32 Win32 https://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20220521-3219d2d-uxp-1e871780f-xpmod-ia32.7z

    source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/ia32

    NM28XP build:
    Win32 https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20220521-d849524bd-uxp-1e871780f-xpmod.7z
    Win32 SSE https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20220521-d849524bd-uxp-1e871780f-xpmod-sse.7z
    Win64 https://o.rthost.win/palemoon/palemoon-28.10.6a1.win64-git-20220521-d849524bd-uxp-1e871780f-xpmod.7z

    Official UXP changes picked since my last build:
    - Issue #1899 - Disable the (broken) MDN integration widget by default. (0749a8828)
    - Issue #1899 - Make sure the test for it still works (349d3d96a)
    - Issue #1898 - Make sure that the sanity test stops running if necessary (d42af42d2)
    - Issue #1813 - Enable date and time picker by default. (40e9f40e8)
    - Issue #1210 - Keep timepicker disabled for now. (8dfcd4553)

    No official Pale-Moon changes picked since my last build.

    No official Basilisk changes picked since my last build.

    My changes picked since my last build:
    - GMP: revert GMPUtils.jsm back to state of rev 4a010b94951a46c3f85da43b1fd712fb1a8f8bc5 (ed4e67920)

    * Notice: From now on, UXP rev will point to `custom` branch of my UXP repo instead of MCP UXP repo, while "official UXP changes" shows only `tracking` branch changes.

    • Like 5

    MyPal 68

    in Browsers working on Older NT-Family OSes

    Posted · Edited by roytam1

    11 hours ago, NotHereToPlayGames said:

    Roytam just compiles them and does not test them

    I have to disagree.

    I do test my builds, but only preliminarily tested. I just can't test every aspects each time I build.

    • Like 6

    My Browser Builds (Part 3)

    in Browsers working on Older NT-Family OSes

    Posted

    12 hours ago, VistaLover said:

    ... I wish I could say the same here (Vista SP2 32-bit)  ;) , however I was bitten by a new regression :( that started with Serpent v52.9.0 (2022-05-06) (32-bit) :angry: ; if you're on WinXP, that one would not be of much impact to you (unless you're using Adobe Primetime CDM as a h264/aac decoder), but if you're using St52 on Vista SP2/Win7 SP1 (or higher...), do stay with me...

    (If you're not interested in the analysis below about EME/DRM and GMPs in Serpent 52, skip that and head to "The regression itself" section)

    First, a "record" of things, well, sort of... As you probably know already, upstream (MCP) harbour an ideological aversion for EME (Encrypted Media Extensions, used in the context of in-browser DRM), as such official Pale Moon never supported it! OTOH, when they created official Basilisk (among other things, to attract "legacy" FxESR52 refugees), they decided to let it stay there, in the state it was inherited from their "upstream", FxESR 52.6.0 .
    EME in the browser entails the installation of two third party DRM plugins, correctly called CDMs (Content Decryption Modules):
    1. Adobe Primetime CDM
    2. Google Widevine CDM
    The first was quickly obsoleted in favour of Google's :angry: one, in fact WV is the sole in-browser CDM used today in desktop browsers to decrypt DRM content...
    Unlike AP, which came with its own patented decoders, WV on a Firefox-type browser relies on OS decoders accessible via WMF (Windows Media Foundation), a Windows feature to be found in fully updated Vista SP2 and higher (i.e., not WinXP) ...
    Google issue frequent updates to their WV CDM, usually to combat discovered and exploited vulnerabilities, but also to remove support for OSes and devices they no longer consider kosher :realmad: , in essence dictating the type of device and client (browser) you can view their DRM'd streams... :realmad:
    The WV CDM is heavily intertwined into the browser's EME implementation, a said Fx version (especially the non-Quantum ones) can only accommodate a specific WV version; FxESR 52 originally came with support for WV v1.4.8.903; when that one was deprecated, MCP tried and managed to equip Basilisk with support for later WV versions, but their effort was forced to stop on (what would turn out to be) final support for WV v1.4.9.1088 ; UXP proved practically "incompatible" with later WV versions (4.9.*, 4.10.*); that very same support has been ported to Serpent 52.9.0 and was present until v52.9.0 (2022-04-29) .

    WV v1.4.9.1088 was deprecated on Aug 14th 2019, since then St52 can't decrypt DRM content when used in Vista SP2/Win7 SP1 :(; FWIW, the latest WV version is v4.10.2449.0, the CDM's dll requires Win7 SP1 or higher, Chrome 69 or higher, FxESR 91 or higher...

    While St52's WV support is now "broken" for decrypting purposes, I still regard it as "working" for, at least, correctly identifying the browser is requested to play back a DRM'd stream; once so, I can then seek playback on another supported device in my household (Win7 and/or Win10 :angry: laptops); e.g. when loading the DRM test case by bitmovin in St52 (2022-04-29), I get this: 

    MzrzUAI.jpg

    i.e. the WV CDM is correctly picked up by the site, and the DRM indicator is displayed in the URL bar, to the left of the padlock; as said, content can't be decrypted, the old CDM has been deprecated and the WV lic servers have blacklisted it... :(

    Unable to progress further in this field, MCP let Basilisk's DRM support rot away, but the official stance from them is Google's refusal to grant "margin" browsers the rights to use their CDM (well, this is true, but you get my drift :P ) ... Thus, upstream no longer check/care for/cater to EME/DRM support in their UXP tree, more so after officially declaring Basilsk as EoL a few days ago... :(

    Another technology MCP feel opposed to is WebRTC (never supported by them in PM), but they also left it in in official Basilisk; in the context of WebRTC, Serpent 52 downloads and installs the "OpenH264 Video Codec provided by Cisco Systems, Inc" plugin, at version 1.7.1; as indicated by MCP, WebRTC support is also to vanish from within their UXP tree...

    The two EME CDMs and the Cisco plugin (not a CDM) are referred to collectively as GMPs (Gecko Media Plugins).

    Disclaimer: When I set out to write up posts like this one, I intend them to serve as sort of Knowledge Bases, "there" even for future reference (as long as MSFN is up) by any interested party, even outside of MSFN; I'm fully aware though, that the length of such posts of mine doesn't bode well with the attention lifespan of many co-members/forum readers, I apologise to them, but they are always free to skip content...

    The regression itself

    Starting with Serpent 52.9.0 (2022-05-06), support for ALL GMPs has been completely BROKEN - this includes both the two CDMs (Adobe Primetime, Google Widevine) and the Cisco Video Codec :( ; I was still on a (2022-04-29) profile myself, when I decided to jump directly on to latest St52 build, (2022-05-12); I did not become aware of the breakage right away, but only when I tried loading a certain DRM stream (the one I troubleshot in my recent MSFN post here) and witnessed the browser acting "odd": the DRM indicator never came up in the URL bar, while it did so in FxESR 52.9.1... Additional troubleshooting revealed in fact that the DRM "breakage" started with the previous St52 release, (2022-05-06) :( ...

    STR

    (OS to be used is Vista SP2 - fully updated to EoS - and higher, XP is NOT suitable!)
    1. Launch a new fresh profile of St52 (2022-04-29) (32-bit), package name is
    "basilisk52-g4.8.win32-git-20220430-3219d2d-uxp-cf4e046f9-xpmod.7z"
    2. Load "about:preferences#content"; you should see the "DRM content" section; tick the "Play DRM content" box:

    HkQQmZS.jpg

    3. Give it 2-5min (YMMV), then load "about:addons => plugins"; you should see entries there for ALL 3 GMPs;

    R75frIr.jpg

    NB: The AP CDM won't be auto-installed "shortly", actually, as the old (internal) download link is no more valid; there's a way to manually install if you have an archived copy of the CDM, but the process is beyond the scope of this bug report...

    4. Exit the browser
    5. Update the browser to Serpent 52.9.0 (2022-05-06) (32-bit), package name is
    "basilisk52-g4.8.win32-git-20220507-3219d2d-uxp-e207b5a16-xpmod.7z"
    6. Launch the browser, it will use by default the previously created profile (by v2022-04-29)
    7. Navigate to "about:addons => plugins"; you'll witness that ALL previous entries of the 3 GMPs have vanished:

    2G2itha.jpg

    8. In a quasi-similar test, delete the existing St52 test profile (with browser closed) and relaunch Serpent 52.9.0 (2022-05-06) (32-bit); a new fresh profile will be created.
    9. In that fresh profile, repeat step [2]; when, afterwards, you repeat step [3], no sign of the 3 GMPs inside the plugin manager...
    10. The next (latest) St52 release, Serpent 52.9.0 (2022-05-12) (32-bit), package name
    basilisk52-g4.8.win32-git-20220514-3219d2d-uxp-774750839-xpmod.7z,
    exhibits the exact same behaviour, i.e. no GMPs are either picked up from an existing profile nor installed in a fresh one:

    0njgjlB.jpg

    Regression range between 2022-04-29 (last GOOD) and 2022-05-06 (first BAD): 

    https://github.com/roytam1/UXP/compare/cf4e046...e207b5a

    (I do see a reference to "gmp" here , but that commit made it to the (2022-05-12) build, also broken... :( )

    Dear @roytam1, no doubt this is all caused by an "upstream" cock-up :angry:, do you think you can restore GMP (and, specifically, WV) support back into Serpent 52? Partial/"broken" as it might be in the (2022-04-29) build, I do have use for it...

    Currently back to the St52 (2022-04-29) release, which, sadly, misses the (?.) upstream implementation...
    Many thanks in advance :) ...

    should be fixed by commit https://github.com/roytam1/UXP/commit/ed4e67920f81c1578a875925efb20188a2281a33

    • Like 2


×
×
  • Create New...