SGI: Development

beta test: firefox3.0.19 - Page 1

I uploaded neko_firefox3.0.19_beta3.tardist to incoming and send a message to nekonoko, it will show up in beta soon. This build is optimized with many of the options shadeofblue suggested and it is noticeable faster than the previous builds. It also fixes a bus error in the svg code and contains updated patch and release notes.

Please report in this thread only severe issues in conjunction with the firefox3 build. I will not fix problems with depending nekoware packages, firefox extensions, your system setup or broken Internet sites.

Next step for me is a complete review of the patch from dexter1, most of his fixes are mips pro related and are not interesting for future builds anymore. The next build will based on a current cvs checkout, so it will be a firefox3.0.19pre.

You can do a mips pro build with the current patch. I did a -O3 build and beside the fact that it had a lot of problems, it was slower than my gcc builds. The beta3 build is much faster now. If anyone is able to make a faster mips pro build compared to this gcc4.7 build I have a free Tezro for him.
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
seems fine on http sites (though only tried a handful so far) but not https - I'm getting the error

Incorrect Response
Firefox doesn't know how to communicate with the server.
The site responded to the network request in an unexpected way and the browser cannot continue.

Perhaps bad preferences on my part?
Fuel ; Indigo2 ; Octane ; RiscPC Kinetic/448MB/RISCOS4.39 or Debian-etch; EspressoPC ViaC3/900MHz/256MB/Debian-testing; RPi B RISCOS5.23; Rpi2 Raspbian-jessie; A5000/33MHz/FPA11/8MB/RISCOS3.11; A540/25MHz/FPA10/16MB/RISCOS3.11 or RISCiX1.21; R140/35MHz/4MB/RISCOS3.11 or RISCiX1.21
ajw99uk wrote: seems fine on http sites (though only tried a handful so far) but not https - I'm getting the error

Incorrect Response
Firefox doesn't know how to communicate with the server.
The site responded to the network request in an unexpected way and the browser cannot continue.

Perhaps bad preferences on my part?

No problem here, I did my online banking with it. Please check your add-ons, noscript? We are talking about beta3?
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
diegel wrote: I uploaded neko_firefox3.0.19_beta3.tardist to incoming

Thank you.

Please report in this thread only severe issues in conjunction with the firefox3 build.

Other people may have some of these same problems. Why waste efforts ?

I will not fix problems with depending nekoware packages,

Again, other people may have installed some of the broken /beta gtk2 neko tardists. If my time can help them, it's fine by me.

firefox extensions,

So far, the only problem I have had with extensions is that you can't get to the secure sites to check them for updates :(

This is even true if you remove the extension totally :( There really is a problem with https://

or broken Internet sites.

When the broken internet site is a major metropolitan newspaper which worked fine (altho displayed strangely) under Firefox 2, perhaps there is a deeper problem. It's worth looking into, at any rate.

Firefox 3 does render pages better than 2. Thank you for that.

however, there is a problem with https:// sites A quick and easy test is to go to duckduckgo.com or ixquick.com Or try to check up ... oops !

Code: Select all

urchin 3% firefox3
moz_run_program[36]: 1351 Memory fault

that didn't work out so well ... did work yesterday tho, so perhaps that was a fluke.

Anyway, secure sites are definitely a problem, at least for me. Turned on ssl2 in about:config, still cannot connect to secure sites. So that is what I would call a for-real difficulty.

Otherwise, it's very nice (now that it looks like an Irix application :D ) And speed-wise, it's quite usable.

thank you.

edit : whatever SFGate does, it still does

Code: Select all

terminate called after throwing an instance of 'std::bad_alloc'
what():  std::bad_alloc
moz_run_program[36]: 1417 Abort

if you open several pages in multiple tabs. I tried deleting XPC.mfasl and letting the app regenerate that file with no success :( This may be a version-specific problem ?
I can't see any ssl problem here. Probably it is a packaging issue, I will test ssl with a fresh installation today. But if I don't see a problem, I am not able to debug it.
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
diegel wrote: I can't see any ssl problem here. Probably it is a packaging issue, I will test ssl with a fresh installation today. But if I don't see a problem, I am not able to debug it.

ajw99uk and I both see this so I don't think it's just here (otherwise I'd agree with you, most likely a problem with my setup.)

Can you connect to ixquick and duckduckgo without problems ? Also the mozilla extensions site refuses to connect.

Looking through about:config, there's a ton of various security settings, many of which are set to false. I could post a listing of those if that would help.

I am using the neko_openssl 0.9.8x , don't know if that is a problem. Does Firefox use an external SSL library ? i didn't see anything about that in the configure settings but ... could there be a clash there ?

Anything you'd like us to try, I would be happy to do.
hamei wrote: Can you connect to ixquick and duckduckgo without problems ? Also the mozilla extensions site refuses to connect.
Yes, both are working fine.
I am using the neko_openssl 0.9.8x , don't know if that is a problem. Does Firefox use an external SSL library ? i didn't see anything about that in the configure settings but ... could there be a clash there ?
firefox uses it's own libssl3.so
Try moving your .mozilla to .mozilla.bak and start firefox3. If the problem still persists it has nothing to do with your firefox configuration.
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
diegel wrote: Try moving your .mozilla to .mozilla.bak and start firefox3.

Oooh nooooo !! My eyes ! My eyes !

Did fix the problem though, thanks very much :D

aj, try a new .mozilla folder ....

Wikipedia disappeared tho. Strange. I don't think it's blocked but the Great Firewall is a strange and terrible creature :shock:

Let's go see if SFGate still blows up the browser ...

thanks much, Mr D. Hurdles were jumped. It wouldn't hurt if a few more people would give it a spin, <cough> <cough> ....
Be aware that . mozilla hosts also your seamonkey, thunderbird and sunbird configuration. The firefox config is in .mozilla/firefox.
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
Thanks for the suggestion, and for the effort put into the package - http pages render nicely, and it seems quicker to start and smoother than FF2.

Changing .mozilla/firefox to .mozilla/firefox.bak : FF3 asked me where I wanted to inherit preferences from, I said none; for https, again got the same error about being unable to communicate with the server.
Changing .mozilla to .mozilla.bak : FF3 started straight into "first time" page; for https, again got the same error about being unable to communicate with the server. This time certainly no add-ons.

The Help->About pop-up shows Mozilla/5.0 (X11; U; IRIX64 IP35; en-US; rv:1.9.0.19) Gecko/2013012608 Firefox/3.0.19, downloaded and installed yesterday evening.

Is there a logfile I should be looking at for clues?
Fuel ; Indigo2 ; Octane ; RiscPC Kinetic/448MB/RISCOS4.39 or Debian-etch; EspressoPC ViaC3/900MHz/256MB/Debian-testing; RPi B RISCOS5.23; Rpi2 Raspbian-jessie; A5000/33MHz/FPA11/8MB/RISCOS3.11; A540/25MHz/FPA10/16MB/RISCOS3.11 or RISCiX1.21; R140/35MHz/4MB/RISCOS3.11 or RISCiX1.21
ajw99uk wrote: Changing .mozilla/firefox to .mozilla/firefox.bak


aj : try this :

start firefox as < firefox3 -profilemanager >

I didn't even let it put the profile in the default place but made a new profile in a new directory entirely.

It is working correctly as of this moment.

Haven't customized anything yet though ....

Report back, please ?

edit : okay, that didn't last long. A new profile seems to work until you close Firefox . When you restart it, the ability to connect to secure sites is gone.

At least that's what it looks like from here.

If I delete every file in the profile folder and make it rebuild the entire thing, then we have secure connections again.

hope this helps.
I was able to reproduce the problem. My package misses the freebl libraries. If you add the firefox2 or seamonkey lib directory to your LD_LIBRARY32_PATH it will work. For example LD_LIBRARY32_PATH=$LD_LIBRARY32_PATH:/usr/nekoware/lib/firefox-2.0.0.22pre. Or copy libfreebl3.chk and libfreebl3.so from /usr/nekoware/lib/firefox-2.0.0.22pre to /usr/nekoware/lib/firefox-3.0.19. I will add the libraries to the next package.
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
There are also two missing dependencies: a current neko_gettext and neko_pixman is needed.
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
I also get the "Firefox doesn't know how to communicate with the server", even after wiping ~/.mozilla, on 2 different systems.

Additionally, beta3 crashes reliably (also on both systems, I have encountered other URLs that crash it, but I'm not sure which they where) when opening http://www.kickstarter.com/projects/wei ... -smbc-coll

Code: Select all

(dbx) run
Process  3334 (firefox-bin) started
Process  3334 (firefox-bin) stopped on signal SIGSEGV: Segmentation violation (default) at [__nanosleep:15 +0x8,0xfabf048]
Source (of /xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/sys/nanosleep.s) not available for Process  3334
(dbx) where thread all

Thread 0x10000
>  0 __nanosleep(0xc098d40, 0x0, 0x1, 0x0, 0x2, 0x400, 0x11a76538, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/sys/nanosleep.s":15, 0xfabf048]
1 libc_lock_lock_mp(0x0, 0x0, 0x1, 0x0, 0x2, 0x400, 0x11a76538, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/libcthread.c":555, 0xc089548]
2 ___libc_lockmalloc(0xc098d40, 0x0, 0x1, 0x0, 0x2, 0x400, 0x11a76538, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/mp/libcthread.c":286, 0xfab1650]
3 _malloc(0x0, 0x0, 0x1, 0x0, 0x2, 0x400, 0x11a76538, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/gen/malloc.c":185, 0xfa444fc]
4 <Unknown>() [< unknown >, 0x49512e4]

Thread 0x10001
>  0 mtx_sys_inc(0x0, 0x4a9bad8, 0x0, 0x1183b910, 0x1, 0x0, 0x0, 0x123e9aa0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/mtx.c":1509, 0xc08bdec]
1 pthread_mutex_lock(0x0, 0x4a9bad8, 0x0, 0x1183b910, 0x34, 0x34, 0x33, 0x112a0fa0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/mtx.c":329, 0xc08a1bc]
2 <Unknown>() [< unknown >, 0x1f2cda8]

Thread 0x10002
>  0 _prctl(0x15, 0x2, 0x200000, 0x11762378, 0x40000, 0x4, 0x68a1, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/proc/prctl.s":15, 0xfa4ac28]
1 _SGIPT_sched_block(0x2, 0x2, 0x200000, 0x11762378, 0x40000, 0x4, 0x68a1, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":763, 0xc0929b0]
2 pthread_cond_timedwait(0x1178ce3c, 0x200ed8, 0x118f3e50, 0x11762378, 0x40000, 0x4, 0x68a1, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/cv.c":576, 0xc08708c]
3 <Unknown>() [< unknown >, 0x1f2d0b0]

Thread 0x20004
>  0 sched_subtract_vp(0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":1025, 0xc0930e0]
1 _SGIPT_sched_block(0x2, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":677, 0xc0927c8]
2 pthread_cond_timedwait(0x11b51cd4, 0x200ed8, 0x13dbbdf0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/cv.c":576, 0xc08708c]
3 <Unknown>() [< unknown >, 0x1f2d0b0]

Thread 0x80000000
>  0 _prctl(0x15, 0x2, 0x11762ba8, 0xc098d58, 0x40000, 0x4, 0x68a1, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/proc/prctl.s":15, 0xfa4ac28]
1 vp_idle() ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":1531, 0xc093f04]

Thread 0x80000003
>  0 __sigpoll(0x119c3f10, 0x119c3f28, 0x119c3f20, 0x58000000, 0x40000, 0x4, 0x68a1, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/signal/sigpoll.s":22, 0xfa426f8]
1 _sigpoll(0x119c3f10, 0x119c3f28, 0x119c3f20, 0x58000000, 0x40000, 0x4, 0x68a1, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/signal/sigpollSCI.c":31, 0xfaf021c]
2 evt_thread(0x119c3f10, 0x119c3f28, 0x119c3f20, 0x58000000, 0x40000, 0x4, 0x68a1, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/event.c":64, 0xc087958]
3 _SGIPT_pt_start() ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/pt.c":813, 0xc08d42c]


Code: Select all

$ dbx /usr/nekoware/lib/firefox-3.0.19/firefox-bin
dbx version 7.3.3 (78517_Dec16 MR) Dec 16 2001 07:45:22
Unable to find libsqlite3.so.1
Unable to find libpng.so
WARNING: Could not find symbol tables for the 2 objects named. Is the environment variable LD_LIBRARY_PATH not set (correctly) or are dsos missing?
Cannot find DSO named libsqlite3.so.1 in current library search path
Cannot find DSO named libpng.so in current library search path
*STRIPPED* Executable /usr/nekoware/lib/firefox-3.0.19/firefox-bin
(dbx) run
Process  1918 (firefox-bin) started
Process  1918 (firefox-bin) stopped on signal SIGSEGV: Segmentation violation (default) at [<unknown>:0 ,0x10ce513c]
(dbx) where thread all

Thread 0x10000
>  0 <Unknown>() [< unknown >, 0x10ce513c]

Thread 0x10001
>  0 mtx_sys_inc(0x0, 0x494bad8, 0x0, 0x11837920, 0x1, 0x0, 0x0, 0x1195bfd8) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/mtx.c":1509, 0xc08bdec]
1 pthread_mutex_lock(0x0, 0x494bad8, 0x0, 0x11837920, 0x34, 0x34, 0x33, 0x4026574) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/mtx.c":329, 0xc08a1bc]
2 <Unknown>() [< unknown >, 0xcacda8]

Thread 0x10002
>  0 _prctl(0x15, 0x2, 0x200000, 0xc098d58, 0x0, 0x1, 0x1, 0x1171d1c0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/proc/prctl.s":15, 0xfa4ac28]
1 _SGIPT_sched_block(0x2, 0x2, 0x200000, 0xc098d58, 0x0, 0x1, 0x1, 0x1171d1c0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":763, 0xc0929b0]
2 pthread_cond_timedwait(0x11789894, 0x200ed8, 0x118cfe50, 0xc098d58, 0x0, 0x1, 0x1, 0x1171d1c0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/cv.c":576, 0xc08708c]
3 <Unknown>() [< unknown >, 0xcad0b0]

Thread 0x20003
>  0 sched_subtract_vp(0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":1025, 0xc0930e0]
1 _SGIPT_sched_block(0x2, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":677, 0xc0927c8]
2 pthread_cond_timedwait(0x1181133c, 0x200ed8, 0x11b1bea0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/cv.c":576, 0xc08708c]
3 <Unknown>() [< unknown >, 0xcad0b0]

Thread 0x30004
>  0 sched_subtract_vp(0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":1025, 0xc0930e0]
1 _SGIPT_sched_block(0x2, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":677, 0xc0927c8]
2 pthread_cond_timedwait(0x11af3b9c, 0x200ed8, 0x11fc3df0, 0x0, 0x0, 0x0, 0x0, 0x0) ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/cv.c":576, 0xc08708c]
3 <Unknown>() [< unknown >, 0xcad0b0]

Thread 0x80000000
>  0 _prctl(0x15, 0x2, 0x11762ba8, 0x11ca9650, 0x1, 0x4, 0x11762378, 0x0) ["/xlv41/6.5.30m/work/irix/lib/libc/libc_n32_M4/proc/prctl.s":15, 0xfa4ac28]
1 vp_idle() ["/xlv41/6.5.30m/work/eoe/lib/libpthread/libpthread_n32_M3/vp.c":1531, 0xc093f04]


When browsing file:///, the Icon for files seems to be missing, they just show the alt image, unlike directories.

While you're adding things to the package, please consider adding a filetype rule and .fti, so that those don't disappear if neko_firefox-2.0 is removed. While you're there, you could automatically add a symlink in the WebTools page of the IconCatalog. This would also be a good opportunity to choose a new tag, since 0x100013 looks suspicous.
diegel wrote: There are also two missing dependencies: a current neko_gettext and neko_pixman is needed.


I have gettext installed from current, and I think I also have pixman. So the necessary files should be there for firefox3. Not at the Fuel just now but will try the copy-over of files from seamonkey / FF2 that you suggested in your previous post.

Having followed suggestions to create a new profile and start with no .mozilla at all, I was still getting the https error - not even getting as far as hamei seemed to. But I could, for example, log into nekochan and post, because it is not https. Will look forward to trying the next iteration (though it may be a couple of days as I'm travelling tomorrow) - thanks again for making this happen.
Fuel ; Indigo2 ; Octane ; RiscPC Kinetic/448MB/RISCOS4.39 or Debian-etch; Dell Inspiron4100/P3 1GHz/1GB/Debian-stable; EspressoPC ViaC3/900MHz/256MB/Debian-testing; RPi B RISCOS5.23; Rpi2 Raspbian-jessie; A5000/33MHz/FPA11/8MB/RISCOS3.11; A540/25MHz/FPA10/16MB/RISCOS3.11 or RISCiX1.21; R140/35MHz/4MB/RISCOS3.11 or RISCiX1.21
diegel wrote: Or copy libfreebl3.chk and libfreebl3.so from /usr/nekoware/lib/firefox-2.0.0.22pre to /usr/nekoware/lib/firefox-3.0.19. I will add the libraries to the next package.

And a restart ... YES !!!

Works well now. Will probably still crash on some URL's but that's certainly not your fault.

If it seems like we are beating you up, I apologize. Really, we're not. I'll go out on a limb and speak for everyone when i saw we all really really really appreciate this. For myself, I can live fine with a lot of older applications but a non-functional browser makes life difficult today. I'll be the first to say FF3 is not very good, but it's a ton better than FF2 ! You've just advanced the shelf life of Irix by several years* :D


* (Which is more than the entire pack of Bozo Ewald clowns in Mountain View did for the last ten years they ran the company. Diegel for CEO !)
One day / two day test update :

Runs pretty well, definitely worth the upgrade. Renders pages much better than FF2. I don't mean artistically or the wonders of cairo, just more accurate interpretation of html.

The secure sites problem is fixed, thank you diegel.

It still crashes but not as much as FF2. Hopefully, now that diegel has broken the barrier, some of these problems can be fixed with newer Flop code. A couple of crashes I get were problems in the past with flop 3 on other platforms ...

Code: Select all

(firefox-bin:7001): Pango-WARNING **: shaping failure, expect ugly output. shape-engine='BasicEngineFc', font='Forefront Book 11.51953125', text='to http://mplayerplug-in.sourceforge.net/download.php.'

and

Code: Select all

terminate called after throwing an instance of 'std::bad_alloc'
what():  std::bad_alloc
moz_run_program[36]: 1948 Abort


The pango error seems to be font-related, might be easy to fix locally. The second one ... code in Firefox itself ? I'm wondering if the MIPSPro code shows the same problem ? Sometimes it doesn't want to refresh the graphics of the webpage ... pita but not mortal.

You can make it look reasonable, too :)

Definintely an improvement. Highly recommended !
Thanks for testing so far. I will upload a new package including the missing files and dependencies on Friday.

@canavan: I don't see the crash you reported. Can you give some Information about your system and your Irix version?
:Tezro: :Fuel: :Octane2: :Octane: :Onyx2: :O2+: :O2: :Indy: :Indigo: :Cube:
canavan wrote: Additionally, beta3 crashes reliably ... when opening http://www.kickstarter.com/projects/wei ... -smbc-coll

Just too a brave pill and tried this, it loaded okay. Took quite a while and then, after everything was loaded it took for eeeeeeever waiting for S3.amazonaws.com then the same for platform.twitter.com and finally facebook.com.

Not a chance in hell those will ever load here, they are blocked. Maybe some of the advertising garbage I have blocked is what crashes the browser for canavan ?

Is there any way to shorten the time Firefox waits for a site before giving up ? I've binged around a little but haven't found anything. Maybe in the code somewhere ? That seems to be one of the problems - while it is sitting there waiting for tab 1 to reply, tabs 2 3 and 4 are on hold.

Also, do you two sometimes have problems with re-drawing the screen ?

diegel : is there a major change between 3.0 and 3.5 ? I know there was a big different at 3.6 but is it possible that 3.5 would solve a few more problems ?