SeaMonkey 2.21b1 and Thunderbird 24.8.1b1

[Dave Yeo]

If anyone is interested I've uploaded SeaMonkey 2.21b1 and Thunderbird 24.8.1b1 to Hobbes.
Based on the Bitwise work for Firefox 24.8.1, same requirements and same issues.
Thanks to Bitwise for their work

[Andi B.]

Thank you Dave. I'm very interested in Seamonkey.

Is there some improvement in speed compared to 2.14? AFAIR 2.14 is the version which is considerable slower than previous ones.

[muffetta]

Hi, Dave, here (on eCS 2.2b2) i got always a crash when i click on a file to download it (no "Download Manager" window for monitoring it), so Ctrl+j ant then the browser crash.
This is the report from POPUPLOG.OS2:

11-03-2014  05:48:09  SYS2070  PID 0084  TID 0001  Slot 0092
C:\DOWNLOADS\SEAMONKEY-2.21B1.EN-US.OS2\SEAMONKEY\SEAMONKEY.EXE
XUL->MOZJS.__Z21JS_GetArrayBufferDataP8JSObject
127
------------------------------------------------------------

11-03-2014  05:48:09  SYS3170  PID 0084  TID 0001  Slot 0092
C:\DOWNLOADS\SEAMONKEY-2.21B1.EN-US.OS2\SEAMONKEY\SEAMONKEY.EXE
c0010001
10825220
EAX=00000000  EBX=fffffff4  ECX=fffffffe  EDX=fffffffc
ESI=12fe6594  EDI=00000000 
DS=0053  DSACC=f0f3  DSLIM=ffffffff 
ES=0053  ESACC=f0f3  ESLIM=ffffffff 
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:1db70173  CSACC=f0df  CSLIM=ffffffff
SS:ESP=0053:0012ff14  SSACC=f0f3  SSLIM=ffffffff
EBP=0012ff70  FLG=00010246

XUL.DLL 0001:007c5220

and this is the running log:
VMI_CMD_SETPTR - rc= 0  ulStatus= 5
Video driver is SNAP - mouse ptr will be hidden
_init_dive:  _load_dive() succeeded
_init_dive:  hDive= 1  scrn= 0x2a50000  format= BGR4
Changing DIVE from enabled to disabled - 'layers.acceleration.disabled' is TRUE
VMI_CMD_SETPTR - rc= 0  ulStatus= 5
Video driver is SNAP - mouse ptr will be hidden
_init_dive:  _load_dive() succeeded
_init_dive:  hDive= 1  scrn= 0x2a40000  format= BGR4
Changing DIVE from enabled to disabled - 'layers.acceleration.disabled' is TRUE
VMI_CMD_SETPTR - rc= 0  ulStatus= 5
Video driver is SNAP - mouse ptr will be hidden
_init_dive:  _load_dive() succeeded
_init_dive:  hDive= 1  scrn= 0x2a40000  format= BGR4
Changing DIVE from enabled to disabled - 'layers.acceleration.disabled' is TRUE
VMI_CMD_SETPTR - rc= 0  ulStatus= 5
Video driver is SNAP - mouse ptr will be hidden
_init_dive:  _load_dive() succeeded
_init_dive:  hDive= 1  scrn= 0x2a50000  format= BGR4
Changing DIVE from enabled to disabled - 'layers.acceleration.disabled' is TRUE
VMI_CMD_SETPTR - rc= 0  ulStatus= 5
Video driver is SNAP - mouse ptr will be hidden
_init_dive:  _load_dive() succeeded
_init_dive:  hDive= 1  scrn= 0x2a40000  format= BGR4
Changing DIVE from enabled to disabled - 'layers.acceleration.disabled' is TRUE
VMI_CMD_SETPTR - rc= 0  ulStatus= 5
Video driver is SNAP - mouse ptr will be hidden
_init_dive:  _load_dive() succeeded
_init_dive:  hDive= 1  scrn= 0x2a60000  format= BGR4
Changing DIVE from enabled to disabled - 'layers.acceleration.disabled' is TRUE
VMI_CMD_SETPTR - rc= 0  ulStatus= 5
Video driver is SNAP - mouse ptr will be hidden
_init_dive:  _load_dive() succeeded
_init_dive:  hDive= 1  scrn= 0x2a50000  format= BGR4
Changing DIVE from enabled to disabled - 'layers.acceleration.disabled' is TRUE

No crash using eCS 2.1GA or any OS/2, maybe the new FOC in eCS 2.2.
Is there a workaround or a solution?

[Dave Yeo]

Thank you Dave. I'm very interested in Seamonkey.

Is there some improvement in speed compared to 2.14? AFAIR 2.14 is the version which is considerable slower than previous ones.

Yes, the speed is much improved over 2.14 and probably close to 2.07.
I see Hobbes seems to have been compromised so have to get the binaries from the OS/2 site

[Dave Yeo]

Hi Barb,
This is a crash I haven't seen before though I haven't tested on 2.2b2 yet due to my dial-up barely working.
I note that you are using SNAP which is totally SMP safe, are you using a multi-core machine? Have you tested Firefox 24.8.1? If not can you test and if you get a similar crash open an issue? Hopefully it'll have the same failure and the bitwise folks would I'm sure be better at debugging this.
Why don't you have a "Download Manager Window"? Mine opens fine here when downloading a file, perhaps profile issue?
And what do you mean by the "new FOC"?
I'll try testing under eCS 2.2b2 later and perhaps do another install enabling SNAP, I take it your install is fairly basic?

[Dave Yeo]

I also just noticed that the IPC plugin supported reverted to enabled. I'd suggest going to about:config and toggling dom.ipc.plugins.enabled to false.
I also had an issue here with SeaMonkey tying up my pipe downloading the safe browsing data and had to go to about;config to turn it off,  toggle these to false,
browser.safebrowsing.enabled
browser.safebrowsing.malware.enabled
services.sync.prefs.sync.browser.safebrowsing.enabled
services.sync.prefs.sync.browser.safebrowsing.malware.enabled

[muffetta]

Dave, it crash also if i open the Download manager window, without necessarily download something (open the browser and then the dm window) ..the last firefox works fine, and also your last seamonkey (2.14), the same profile for both releases (multi-core machine, not yet tested on my old ibm T23, runnint also eCS 2.2b2).
FOC, because sometimes the dialog boxes (often after a long browsing session) does not open, and I have to restart the browser to get it work again (but this issue there was with all sm releases, not with FOC disabled).

[muffetta]

Ok, now it works: clear the list of downloaded files (too long) into the Download Manager window using seamonkey 2.14 solves the crash, obviously the previous version does not have this issue.

[Andreas Schnellbacher]

FOC, because sometimes the dialog boxes (often after a long browsing session) does not open, and I have to restart the browser to get it work again

Then it looks to me like a Shared Memory problem. The simple way to avoid this is to close Mozilla apps (as well as OpenOffice and Java apps) from time to time.

Andreas

[Dave Yeo]

Yes, I also think a shared memory issue. SeaMonkey should really be marked to load all its DLLs high.
At one time we had a severe performance hit if the number of downloaded files became too big in the download manager and it seems to still be a slight issue, so probably a good idea to prune it now and again.

[Andreas Schnellbacher]

SeaMonkey should really be marked to load all its DLLs high.

It depends if it gives an improvement: You need the SMP kernel with the patch from Holger Veith (14.106) to avoid the hang on DLL unloading.

As a resolution I see the following: To provide a single package only, allow DLL patching on a user's system (both forth and back) and either link or ship the scripts and executable. An option would be to create a separate package for patching and to refer to it in the readme.

[Dave Yeo]

I started to get a similar issue except it crashed in pmmerge and cascaded through a bunch of other system DLLs which I also fixed by emptying the download manager. in my case it crashed when starting to download.
I had lots of shared memory available, about 240MBs and may well be a bug in the widget code somewhere 
I guess eventually I'll have to attempt to build a debug version and try to track it down.
I'd advise everyone to keep the list of downloaded files pruned.

[Doug Bissett]

You need the SMP kernel with the patch from Holger Veith (14.106) to avoid the hang on DLL unloading.

Just to clarify: Does the 14.106 kernel already have the patch, or does it need it?

I am seeing Firefox hang the desktop when it closes, every once in a while. I haven't really figured out what triggers the problem, but I am thinking that it has something to do with viewing PDF files. It is only recently (since the 24 versions, I think) that PDF files have started to display, without messing around with other programs. At least that part seems to work okay.

[Andreas Schnellbacher]

You need the SMP kernel with the patch from Holger Veith (14.106) to avoid the hang on DLL unloading.

Just to clarify: Does the 14.106 kernel already have the patch, or does it need it?

Holger wrote (AFAIR) that it should have. Many years after Holger had sent it to Mensys, Steven most likely included it in 14.106. Steven is the one to ask.

[Andreas Schnellbacher]

Beside the missing default setting

Code: [Select]

dom.ipc.plugins.enabled;false
(thanks Dave for posting this), I have no sound [1] with SeaMonkey 2.21b1.

It works with 2.14. I have created a new profile for to check. Does anyone with Firefox 24 have the same behavior?

[1]  A good example, the weekly Heise podcast: http://www.heise.de/ct/blog-ct-uplink-2176079.html