Author Topic: Experimental build of SeaMonkey  (Read 60054 times)

Roderick Klein

  • Hero Member
  • *****
  • Posts: 655
  • Karma: +14/-0
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #15 on: November 28, 2017, 11:15:58 pm »
I've uploaded an experimental build of SeaMonkey 2.42.9ESR, https://bitbucket.org/dryeo/dry-comm-esr31/downloads/seamonkey-2.42.9esr.en-US.os2_exp1.zip
Optimized for size instead of performance, should fit into memory a bit better and almost as fast. Also optimized for a Pentium M so needs a CPU with SSE2 support.
You could also install it as if it was an RPM, putting everything in @UNIXROOT/usr/lib/seamonkey-2.42.9esr and a copy of seamonkey.exe in @UNIXROOT/usr/bin so it is on the PATH.
Same codebase as the latest release of Firefox. Once dmik releases another version of Firefox, I'll release the usual i686 build and if I ever get some guidance on building an RPM, I'll try to release them as well.
Needs the latest libcx and libc that are now available on netlabs-release RPM repository (I hope).

Note there also seems to be a regression with highmem support, marking xul.dll to load code and data makes SM unstable here on ArcaOS.

In my experience only load the code segments high, not the data segments!

I am running Firefox, Thunderbird and Open Office next to each other without an issue. Without loading the DLL's in high memory it toasts my system within a few minutes.

Roderick

Andreas Schnellbacher

  • Hero Member
  • *****
  • Posts: 827
  • Karma: +14/-0
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #16 on: November 29, 2017, 02:05:38 am »
Note there also seems to be a regression with highmem support, marking xul.dll to load code and data makes SM unstable here on ArcaOS.
In my experience only load the code segments high, not the data segments!
Dmitriy's comments explain what Dave has stated.

(I myself haven't tested Dave's build yet. About loading data segments high: It depends. Unfortunately that would be more effective than code segments in the respect of saving low shared memory.)

Dave Yeo

  • Hero Member
  • *****
  • Posts: 4787
  • Karma: +99/-1
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #17 on: November 29, 2017, 02:27:30 am »
Note there also seems to be a regression with highmem support, marking xul.dll to load code and data makes SM unstable here on ArcaOS.
In my experience only load the code segments high, not the data segments!
Dmitriy's comments explain what Dave has stated.

(I myself haven't tested Dave's build yet. About loading data segments high: It depends. Unfortunately that would be more effective than code segments in the respect of saving low shared memory.)

I think it is a different issue that Dmitriy is talking about, namely dynamically allocating memory high.
Here SM 2.35 is stable with the DLLs marked to load both data and code high and currently I have Thunderbird-2.4.8 DLLs marked to load high while SeaMonkeys are not marked and the system is fairly stable, at least for a day or two.

Dave Yeo

  • Hero Member
  • *****
  • Posts: 4787
  • Karma: +99/-1
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #18 on: November 29, 2017, 02:38:29 am »
My libpath in config.sys:
LIBPATH=.;C:\usr\local\lib;C:\usr\lib;C:\usr\lib\unixroot\usr\lib\firefox-45.9.0;C:\PROGRAMS\NETSCAPE\PROGRAM;C:\SYS\DLL;etc etc but not the folder.
In this libpath I have the xull.dll modify with highmen
but running this other firefox45.9 in a folder, you can see the trp, because after send me previous message hang with this trp.
In popup.os2 i have not any about firefox.
You can see that is run in d:\util2017test\....
Do you want me to include this folder in libpath? and wait for it to crack, and take out the trp?
saludos

You should remove C:\usr\lib\unixroot\usr\lib\firefox-45.9.0 from your LIBPATH and run C:\usr\lib\unixroot\usr\lib\firefox-45.9.0\firefox.exe Any program objects should also use this as the working directory.

Andreas Schnellbacher

  • Hero Member
  • *****
  • Posts: 827
  • Karma: +14/-0
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #19 on: November 29, 2017, 02:44:48 am »
I think it is a different issue that Dmitriy is talking about, namely dynamically allocating memory high.
OK, with your explanation I see the difference, but I don't fully understand memory allocation in C.

Here SM 2.35 is stable with the DLLs marked to load both data and code high
Here as well.

xynixme

  • Guest
Re: Experimental build of SeaMonkey
« Reply #20 on: November 29, 2017, 05:15:41 pm »
In this libpath I have the xull.dll modify with highmen

</silence>
FWIW: are you using SET LIBPATHSTRICT=T and e.g. SET BEGINLIBPATH=y:\TEST;x:\yz\SEAMONKEY as (hidden) WPS settings for your SM test objects?

If so, then you may not be using the expected LIBPATH directory which contains a patched XUL.DLL.
If not, then you could consider to start using those settings to make sure that FF/SM will use expected DLLs, like a y:\TEST\XUL.DLL or a JPEG.DLL in SM's directory.
<silence>

roberto

  • Hero Member
  • *****
  • Posts: 810
  • Karma: +3/-6
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #21 on: November 29, 2017, 05:26:50 pm »
You should remove C:\usr\lib\unixroot\usr\lib\firefox-45.9.0 from your LIBPATH and run C:\usr\lib\unixroot\usr\lib\firefox-45.9.0\firefox.exe Any program objects should also use this as the working directory.
Finally I have been able to start the seamonkey.
I have removed everything from the libpath. Both in config.sys and in usr \ lib ..
In firefox45.9 I started it from any folder. Unzip and execute, perfect.
But the seamonkey failed again and again, with the path without the path and in usr \ lib. Delete everything in d:\home\seamonkey, copy the profiles from firefox to seamonquey and everything perfect, now I execute it from any folder.
Do not copy the thunderbird, because it's not my idea to use the seamonkey.
The videos look just as bad.
saludos
regards

Pete

  • Hero Member
  • *****
  • Posts: 1281
  • Karma: +9/-0
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #22 on: November 29, 2017, 05:33:21 pm »
Hi Dave

This build works about the same as the previous build: no video playback - cannot even load youtube pages completely. Less of a problem with cpus peaking when using the browser but email/newsgroups causes cpu peaking.

Something to point out from the Readme:-

"The list of RPM packages required by Firefox is
   contained in a file named RPM_REQUIREMENTS inside the ZIP archive."

So it is now necessary to download and unzip the Firefox package to read that readme file to discover what support files are required for Seamonkey!


Regards

Pete


Dave Yeo

  • Hero Member
  • *****
  • Posts: 4787
  • Karma: +99/-1
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #23 on: November 29, 2017, 06:05:08 pm »
Something to point out from the Readme:-

"The list of RPM packages required by Firefox is
   contained in a file named RPM_REQUIREMENTS inside the ZIP archive."

So it is now necessary to download and unzip the Firefox package to read that readme file to discover what support files are required for Seamonkey!

Missed that. If someone wants to send me the file, I can add it to the zip.
At some point hopefully someone will tell me how to create a RPM and then I can generate it myself.

As for the high CPU and failing youtube, hopefully Bitwise will figure it out as it seems to be a problem with the JavaScript engine.

Pete

  • Hero Member
  • *****
  • Posts: 1281
  • Karma: +9/-0
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #24 on: November 29, 2017, 06:39:01 pm »
Hi Dave

Forgot to add that this build of Seamonkey often hangs when starting and can take 3 or 4 attempts to actually get it running.

Maybe I do not have the latest something "lib-nixy" installed but ANPM tells me that everything is up to date...


Regards

Pete



roberto

  • Hero Member
  • *****
  • Posts: 810
  • Karma: +3/-6
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #25 on: November 29, 2017, 08:01:35 pm »
As for the high CPU and failing youtube, hopefully Bitwise will figure it out as it seems to be a problem with the JavaScript engine.
I disagree with this. Today in firefox 45.9. I have been able to watch videos on YouTube as always, even the same mistakes as in previous versions. But I have not been able to repeat it, for now.
When I created a new profile in seamonkey, I stopped booting. It gives me the feeling that the seamonkey, fails in the creation of the profiles, and that's why I could not execute it.
I even think that the seamonkey, I modified the profile of firefox, and that's why I could see the videos well.
saludos

Dariusz Piatkowski

  • Hero Member
  • *****
  • Posts: 1317
  • Karma: +26/-0
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #26 on: November 30, 2017, 04:06:09 am »
Dave!

Something to point out from the Readme:-

"The list of RPM packages required by Firefox is
   contained in a file named RPM_REQUIREMENTS inside the ZIP archive."

So it is now necessary to download and unzip the Firefox package to read that readme file to discover what support files are required for Seamonkey!

Missed that. If someone wants to send me the file, I can add it to the zip....

Attached to the response...taken from the 7z build of Firefox.

Dave Yeo

  • Hero Member
  • *****
  • Posts: 4787
  • Karma: +99/-1
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #27 on: November 30, 2017, 05:13:06 am »
As for the high CPU and failing youtube, hopefully Bitwise will figure it out as it seems to be a problem with the JavaScript engine.
I disagree with this. Today in firefox 45.9. I have been able to watch videos on YouTube as always, even the same mistakes as in previous versions. But I have not been able to repeat it, for now.
When I created a new profile in seamonkey, I stopped booting. It gives me the feeling that the seamonkey, fails in the creation of the profiles, and that's why I could not execute it.
I even think that the seamonkey, I modified the profile of firefox, and that's why I could see the videos well.
saludos

Just tested creating a brand new profile, seemed fine. The big difference with SM from FF is that SeaMonkey installs some extensions by default. Chatzilla for one seems to create a lot of CPU usage and should probably be disabled, as do some others such as adblockplus on Firefox and perhaps SeaMonkey.
Here the only time that SeaMonkey won't start is if I already have Thunderbird running. The other way works fine.
For both you and Pete, does starting in safe-mode help, both starting and playing youtube videos?
Code: [Select]
seamonkey.exe -safe-mode

Dave Yeo

  • Hero Member
  • *****
  • Posts: 4787
  • Karma: +99/-1
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #28 on: November 30, 2017, 05:14:02 am »
Attached to the response...taken from the 7z build of Firefox.

Thanks

roberto

  • Hero Member
  • *****
  • Posts: 810
  • Karma: +3/-6
    • View Profile
Re: Experimental build of SeaMonkey
« Reply #29 on: November 30, 2017, 03:59:42 pm »
For both you and Pete, does starting in safe-mode help, both starting and playing youtube videos?
Code: [Select]
seamonkey.exe -safe-mode
This does not improve anything in my case.

When I create a new profile in seamonkey, 1,2,3, ...
If I modify in firefox \ profiles.ini
StartWithLastProfile=1    replace by 1, 2, or 3 ...
 This allows me to run only one session in firefox 45.9 with the videos ok. I understand that this session does not close well, but that I have to kill it. It does not allow me to repeat it again. Unless you create a new profile ....
saludos