Jump to content

My Browser Builds (Part 1)


Recommended Posts

1 hour ago, i430VX said:

From my laptop:
In new moon and basilisk, the browser freezes up (consuming 50% cpu during this) on ebay item pages and the amazon login screen. It does not seem to matter whether or not "use hardware acceleration" is on, and I have tried every GPU driver version available for my ATi mobility radeon x1300 with no difference.

However these pages work fine on my other computers.

:dubbio:

does it start in browser's safe mode?

Link to comment
Share on other sites


New build of Serpent/UXP for XP!

Test binary:
Win32 https://o.rths.cf/basilisk/basilisk52-g4.1.win32-git-20190323-0d9f3396a-xpmod.7z
Win64 https://o.rths.cf/basilisk/basilisk52-g4.1.win64-git-20190323-0d9f3396a-xpmod.7z

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

NM28XP build:
Win32 https://o.rths.cf/palemoon/palemoon-28.5.0a1.win32-git-20190323-0d9f3396a-xpmod.7z
Win64 https://o.rths.cf/palemoon/palemoon-28.5.0a1.win64-git-20190323-0d9f3396a-xpmod.7z

Official repo changes since my last build:
- [EME] Make WidevineAdapter compatible with CDM version 9 (1fa998d89)
- [EME} Hackily implement cdm::Host_9::RequestStorageId (6310310e0)
- [EME] Update content_decryption_module.h to more recent revision (34bfc1f40)
- [EME] Add content_decryption_module_ext.h (8279f15d4)
- [EME] Cherry-pick fix for content_decryption_module.h (e1d7634ba)
- Style fixes (3a6a37ca1)
- Build bustage fixes (05a51611b)
- [EME] Remove WidevineDecryptor::OnLegacySessionError (cece1cb71)
- Draw a border when hiding the tab bar (ca1417782)
- Don't modify toolbox-after-color in lwthemes (07122c445)
- Fix build bustage. (e9261d329)
- Merge pull request #1011 from Lootyhoof/issue-725 (b0c7cca30)
- Merge branch 'cdm-work' (c7c3c8600)
- Merge branch 'master' of https://github.com/MoonchildProductions/UXP (a36c3f69c)
- Remove harmless assertions by using IgnoredErrorResult. (85e33523c)
- Pref the use of unboxed plain objects in JS and disable by default. (890bb438b)
- Force expected crashes on unexpected magic values. (ff9877cfe)
- Check for dead wrappers in CallerGetterImpl/CallerSetterImpl. (a7013083d)
- Check child container count in NotifyRefreshDriverCreated. (403c0b870)
- Clean up nsOSHelperAppService. (3c02dbad9)
- Extend origin-clean algorithm (DiD). (8549c7cc3)
- Reset nursery position when it's disabled. (c53602c80)
- Guard against re-entrancy in nsStringStream. (411919cca)
- Revert "Guard against re-entrancy in nsStringStream." (34939b2d1)
- Close the transaction if PR_Read/PR_Write failed. (3403d5f04)
- Add SSUAO for polyfill.io (8ca6744b2)
- Return proper error if the nss layer encounters an error on the http tunnel. (a40132518)
- Use ptrdiff_t in RangedPtr. (0d9f3396a)

Link to comment
Share on other sites

New build of BOC/UXP for XP!

Test binary:
MailNews Win32 https://o.rths.cf/boc-uxp/mailnews.win32-20190323-5f8453b-uxp-0d9f3396a-xpmod.7z
Browser-only Suite Win32 https://o.rths.cf/boc-uxp/bnavigator.win32-20190323-5f8453b-uxp-0d9f3396a-xpmod.7z

source patch (excluding UXP): https://o.rths.cf/boc-uxp/boc-uxp-src-xpmod-20190223.7z

Official repo changes since my last build:
- [PLATFORM] Update commit pointer (a3cbf79)
- [PLATFORM] Update commit pointer (b85f3c5)
- [PLATFORM] Update commit pointer (f7996df)
- [MAIL] Always build the calendar (3c17df9)
- [MAIL] Calendar changes for primary Interlink operation (db8a1d6)
- [MAIL] Remove Calendar in-line styling (ca8063c)
- [MAIL] Make the header toolbar visible by pref (5f8453b)

For UXP changes please see above.

Edited by roytam1
Link to comment
Share on other sites

New regular/weekly KM-Goanna release:
https://o.rths.cf/kmeleon/KM76.1.1-Goanna-20190323.7z

Changelog:

Out-of-tree changes:
* update Goanna3 to git 69004b612..fcb2b3aa1:
- import changes from rmottola/Arctic-Fox:
 - Bug 1135141 - Release relocated arenas immediately on last ditch GC (e5053c077)
 - Bug 1135141 - Fix jsapi-test framework to not overwrite original global when createGlobal() fails (055caeee8) (d17844d0f)
- import changes from rmottola/Arctic-Fox:
 - change pointer style (08ae3786b)
 - Bug 1138538 - Display GC zeal help text in shell help message (6c25ae0a8)
 - Bug 1135723 - Traverse layout properties instead of object group properties in makeNativeGroup, handle int32->double coercion in ensureDefiniteType, (e53d9bfe1)
 - Bug 1135535 part 4 - Fix C4275 dll-interface warning and mark this warning as error in js/src on MSVC. (ed0660603)
 - change pointer style (31d69f5d9)
 - Bug 1135985 - Use typed members to implement pushMarkStack (42c197cbe)
 - Bug 1137978 - Access an object's compartment and zone via its group, remove ObjectGroup::singleton_ (bcc16c922)
 - Bug 1137497 - Remove shape from unboxed objects (eb874ad5a)
 - Bug 1138735 - Fallback to software vsync if CVDisplayLink fails to initialize. (40479d546)
 - Bug 1136896. Speed up fill() and dedent() by memoizing some of the work they currently end up doing on each call. (e7d60964c)
 - add header include (d7f3b53ac)
 - Bug 1134425 - Part 1: move the allocator interface code out of line (73a0b7e67)
 - change pointer style for better patching (7ff55e7a5)
 - Bug 1134425 - Part 2: templatize the GC's allocation routines (10f091bb5)
 - Bug 1136345. Drop the parent arg from JS_NewObjectWithGivenProto and introduce a JS_DeprecatedNewObjectWithGivenProtoAndParent for the few cases that still pass in a custom parent. (ebff19bb0)
 - Bug 1136520. Stop parenting Xray expando objects to a non-global. (d58a73a3d)
 - Bug 1134970 part 1. Add JS friend API to allocate an object which is guaranteed to have no attached metadata. (3e8fa3b3e)
 - Bug 1134970 part 2. Add JS friend API to quickly copy properties from one object to another if the objects are similar enough. (a53538b91)
 - Bug 1138874 - Change ReparentWrapper() to avoid multiple JS objects pointing to the same native (16224aa3d)
 - Bug 1138874 - Generalise assertion that source object has a preserved wrapper (bff647e8a) (f66a5071e)
- import changes from rmottola/Arctic-Fox:
 - Bug 1131805 part 1. Remove remaining JS_GetParent and js::GetObjectParent uses in Gecko. (5ec206e98)
 - Bug 1137578. Remove JS_GetParent usage from mozJSSubscriptLoader. (ba8b18d88)
 - Bug 1131802 part 1. Add JS friend API to test whether a given function object has extended slots. (d7a2bad97)
 - Bug 1131802 part 2. Allocate functions with reserved slots for DOM Xrays so we can store the Xray wrapper reference in those slots instead of as the function parent. (7bc6dba71)
 - Bug 1140399. Add friend API to check whether a given function object has a JSNative and to get that JSNative as needed. (71f0ccff7)
 - Bug 1137334. Remove nsIDOMWindowUtils.getParent. (aa3de0a27)
 - Bug 1125784 - Remove JS_ConvertArguments and JS_ConvertValue. (db6a1bd29)
 - Bug 1137325. Remove the parent() function from xpcshell. (7cac0999e)
 - Bug 1131797 part 1. Store an index into its XPCNativeInterface mMembers array in each XPCNativeMember. (2389614e2)
 - Bug 1131797 part 2. Stop storing the XPCNativeInterface* in XPConnect function objects and just get it from the XPCNativeMember* we're still storing. (252172d87)
 - Bug 1131797 part 3. Store the object we originally come from in a reserved slot on XPConnect functions instead of using the parent pointer. (252bb312f)
 and fix requireAtLeast prototype for --enable-shared-js (c660cd77b)
- import changes from rmottola/Arctic-Fox:
 - Bug 1136516. Stop using JS_DeprecatedNewObjectWithGivenProtoAndParent in XPCWrappedNative. (627d0c40f)
 - Bug 1136523. Stop doing weird things with parents in ctypes code and remove the JS_DeprecatedNewObjectWithGivenProtoAndParent API. (075a3be27)
 - Bug 1131805 part 2. Remove remaining js::GetObjectParent and JS_GetParent uses in SpiderMonkey. (6caf6189b)
 - Bug 928336. Make defining unforgeable properties on objects faster by just copying them from an unforgeable holder object. (e7d32668d)
 - Bug 1132522, part 1 - Treat false return from proxyHandler.defineProperty() as strict mode failure. (3237f9b70)
 - Bug 1132522, part 2 - Treat false return from proxyHandler.set() as strict mode failure. (e4ab29f7a) (fcb2b3aa1)
* updated kmeleon76 goanna tree with additional change:
- ported "Allow .js preference files to set locked prefs with lock_pref()" patch (491cdc53d)

* 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

Link to comment
Share on other sites

On 3/23/2019 at 12:31 AM, roytam1 said:

does it start in browser's safe mode?

Yeah it does start in the browsers safe mode. but i have the same issue of it freezing on ebay item pages and amazon login.

I tried the latest builds, too.
Eventually it does unfreeze but not before a few minutes has passed.

Link to comment
Share on other sites

8 minutes ago, i430VX said:

Yeah it does start in the browsers safe mode. but i have the same issue of it freezing on ebay item pages and amazon login.

I tried the latest builds, too.
Eventually it does unfreeze but not before a few minutes has passed.

maybe it is not related to graphics but heavy javascript?

BTW you can try toggling layer.* switches (like layers.prefer-opengl, layers.acceleration.disabled, layers.deaa.enabled, layers.bench.enabled, etc.) in about:config?filter=layer to see if issue can be relexed or not.

Link to comment
Share on other sites

On 3/23/2019 at 1:23 AM, roytam1 said:

maybe it is not related to graphics but heavy javascript?

BTW you can try toggling layer.* switches (like layers.prefer-opengl, layers.acceleration.disabled, layers.deaa.enabled, layers.bench.enabled, etc.) in about:config?filter=layer to see if issue can be relexed or not.

yeah i did notice a couple long-running script errors. what do you recommend to do so i can get them in check?
UPDATE: I tried noscript, seems to do the trick. :yes:

Edited by i430VX
Link to comment
Share on other sites

1 minute ago, i430VX said:

yeah i did notice a couple long-running script errors. what do you recommend to do so i can get them in check?

you may try creating a new clean profile and test sites on that profile (to eliminate issues from extensions)

but if your CPU is too weak/old, then it is not much you can do.

Link to comment
Share on other sites

On 3/23/2019 at 1:31 AM, roytam1 said:

you may try creating a new clean profile and test sites on that profile (to eliminate issues from extensions)

but if your CPU is too weak/old, then it is not much you can do.

Yeah it is fairly old (Core duo 1.83ghz)
Like i included into my edit of the previous post, adding noscript to the browsers has resolved the issue. :)

Link to comment
Share on other sites

New New Moon 27 Build!

32bit https://o.rths.cf/palemoon/palemoon-27.9.6.win32-git-20190323-fcb2b3aa1-xpmod.7z
32bit SSE https://o.rths.cf/palemoon/palemoon-27.9.6.win32-git-20190323-fcb2b3aa1-xpmod-sse.7z
32bit noSSE https://o.rths.cf/palemoon/palemoon-27.9.6.win32-git-20190323-fcb2b3aa1-xpmod-ia32.7z

64bit https://o.rths.cf/palemoon/palemoon-27.9.6.win64-git-20190323-fcb2b3aa1-xpmod.7z

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

repo changes since my last build:
- import changes from rmottola/Arctic-Fox:
 - Bug 1135141 - Release relocated arenas immediately on last ditch GC (e5053c077)
 - Bug 1135141 - Fix jsapi-test framework to not overwrite original global when createGlobal() fails (055caeee8) (d17844d0f)
- import changes from rmottola/Arctic-Fox:
 - change pointer style (08ae3786b)
 - Bug 1138538 - Display GC zeal help text in shell help message (6c25ae0a8)
 - Bug 1135723 - Traverse layout properties instead of object group properties in makeNativeGroup, handle int32->double coercion in ensureDefiniteType, (e53d9bfe1)
 - Bug 1135535 part 4 - Fix C4275 dll-interface warning and mark this warning as error in js/src on MSVC. (ed0660603)
 - change pointer style (31d69f5d9)
 - Bug 1135985 - Use typed members to implement pushMarkStack (42c197cbe)
 - Bug 1137978 - Access an object's compartment and zone via its group, remove ObjectGroup::singleton_ (bcc16c922)
 - Bug 1137497 - Remove shape from unboxed objects (eb874ad5a)
 - Bug 1138735 - Fallback to software vsync if CVDisplayLink fails to initialize. (40479d546)
 - Bug 1136896. Speed up fill() and dedent() by memoizing some of the work they currently end up doing on each call. (e7d60964c)
 - add header include (d7f3b53ac)
 - Bug 1134425 - Part 1: move the allocator interface code out of line (73a0b7e67)
 - change pointer style for better patching (7ff55e7a5)
 - Bug 1134425 - Part 2: templatize the GC's allocation routines (10f091bb5)
 - Bug 1136345. Drop the parent arg from JS_NewObjectWithGivenProto and introduce a JS_DeprecatedNewObjectWithGivenProtoAndParent for the few cases that still pass in a custom parent. (ebff19bb0)
 - Bug 1136520. Stop parenting Xray expando objects to a non-global. (d58a73a3d)
 - Bug 1134970 part 1. Add JS friend API to allocate an object which is guaranteed to have no attached metadata. (3e8fa3b3e)
 - Bug 1134970 part 2. Add JS friend API to quickly copy properties from one object to another if the objects are similar enough. (a53538b91)
 - Bug 1138874 - Change ReparentWrapper() to avoid multiple JS objects pointing to the same native (16224aa3d)
 - Bug 1138874 - Generalise assertion that source object has a preserved wrapper (bff647e8a) (f66a5071e)
- import changes from rmottola/Arctic-Fox:
 - Bug 1131805 part 1. Remove remaining JS_GetParent and js::GetObjectParent uses in Gecko. (5ec206e98)
 - Bug 1137578. Remove JS_GetParent usage from mozJSSubscriptLoader. (ba8b18d88)
 - Bug 1131802 part 1. Add JS friend API to test whether a given function object has extended slots. (d7a2bad97)
 - Bug 1131802 part 2. Allocate functions with reserved slots for DOM Xrays so we can store the Xray wrapper reference in those slots instead of as the function parent. (7bc6dba71)
 - Bug 1140399. Add friend API to check whether a given function object has a JSNative and to get that JSNative as needed. (71f0ccff7)
 - Bug 1137334. Remove nsIDOMWindowUtils.getParent. (aa3de0a27)
 - Bug 1125784 - Remove JS_ConvertArguments and JS_ConvertValue. (db6a1bd29)
 - Bug 1137325. Remove the parent() function from xpcshell. (7cac0999e)
 - Bug 1131797 part 1. Store an index into its XPCNativeInterface mMembers array in each XPCNativeMember. (2389614e2)
 - Bug 1131797 part 2. Stop storing the XPCNativeInterface* in XPConnect function objects and just get it from the XPCNativeMember* we're still storing. (252172d87)
 - Bug 1131797 part 3. Store the object we originally come from in a reserved slot on XPConnect functions instead of using the parent pointer. (252bb312f)
 and fix requireAtLeast prototype for --enable-shared-js (c660cd77b)
- import changes from rmottola/Arctic-Fox:
 - Bug 1136516. Stop using JS_DeprecatedNewObjectWithGivenProtoAndParent in XPCWrappedNative. (627d0c40f)
 - Bug 1136523. Stop doing weird things with parents in ctypes code and remove the JS_DeprecatedNewObjectWithGivenProtoAndParent API. (075a3be27)
 - Bug 1131805 part 2. Remove remaining js::GetObjectParent and JS_GetParent uses in SpiderMonkey. (6caf6189b)
 - Bug 928336. Make defining unforgeable properties on objects faster by just copying them from an unforgeable holder object. (e7d32668d)
 - Bug 1132522, part 1 - Treat false return from proxyHandler.defineProperty() as strict mode failure. (3237f9b70)
 - Bug 1132522, part 2 - Treat false return from proxyHandler.set() as strict mode failure. (e4ab29f7a) (fcb2b3aa1)
 

Edited by roytam1
Link to comment
Share on other sites

15 hours ago, roytam1 said:

Roy, will any of these run on an XP system without a SSE processor? I ask only because it isn't explicitly  mentioned as was the case with your palemoon 27 build that I'm currently using (Thank you very much!). I am using an amd athlon 1200 Mhz. I also have a win 2k sp4 system but I only use it occasionally.

Please excuse my ignorance.

And thanks for the great work you do!

 

Link to comment
Share on other sites

15 hours ago, i430VX said:

yeah i did notice a couple long-running script errors. what do you recommend to do so i can get them in check?
UPDATE: I tried noscript, seems to do the trick. :yes:

I've had that problem with seamonkey for YEARS. And I've used noscript for a very long time. It never helped, but I did allow some scripts for ebay to work. If I really want to buy something badly, I can persist and eventually get it to work, but very slow with all the hangups.  I can't run any seamonkey version later than 2.40 though, because of my old  amd processor (has the curse of no magic sse stuff:)

Could you say what script sources you allowed, if any?

Thanks!

Link to comment
Share on other sites

On 3/23/2019 at 10:36 PM, Old Man said:

will any of these run on an XP system without a SSE processor? I ask only because it isn't explicitly  mentioned as was the case with your palemoon 27 build

Both Serpent 52.9.0 and New Moon 28 (currently in version 28.5.0a1) are built on the UXP application platform (and are forks of official Basilisk 52.9.20xx.xx.xx and Pale Moon [unstable] 28.5.0a1, respectively...).  See: 

System requirements for UXP applications (Win)

https://www.basilisk-browser.org/requirements.shtml

http://www.palemoon.org/download.shtml => 

Quote

Minimum System Requirements:

A processor with SSE2 instruction support

UXP platform's Javascript engine is such that it issues SSE2 instructions irrespective of compiler configuration; this is why NM28+St52 cannot be compiled for even SSE-only CPUs :( ... 

TL:DR: You are confined to your no-SSE NM27 builds... ;)

Edited by VistaLover
Make links live / formatting
Link to comment
Share on other sites

15 minutes ago, VistaLover said:

Both Serpent 52.9.0 and New Moon 28 (currently in version 28.5.0a1) are built on the UXP application platform (and are forks of official Basilisk 52.9.20xx.xx.xx and Pale Moon [unstable] 28.5.0a1, respectively...).  See: 

https://forum.palemoon.org/viewtopic.php?p=142274#p142274

https://www.basilisk-browser.org/requirements.shtml

http://www.palemoon.org/download.shtml => 

UXP platform's Javascript engine is such that it emits SSE2 instructions irrespective of compiler configuration; this is why NM28+St52 cannot be compiled for even SSE-only CPUs :( ... 

TL:DR: You are confined to your no-SSE NM27 builds... ;)

Thanks, that's what I figured, but asked anyway. Its the old AMD curse. Henceforth I shall buy only Apple, intel and microsoft products. I've learned my lesson. I hope they make good underwear. Mine are threadbare :)

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...