"http://wwww.linkitup.de/ForAIX/Firefox"
Firefox 2.0 has grown and is quite a lot slower than 1.5.0.7 and 1.5.0.8rc2
I also have Seamonkey 1.0.5 for those who like Mozilla 1.7
--
Uli Link
If you will pardon the pun, I cannot get your Link to work.....
CT the Inept
Uli suffered from a RTO (repetitive typing error) : one W too much
Try : www.linkitup.de/ForAIX/Firefox
Have fun!
"Tim Knight/CT" <msgtk...@gmail.com> schreef in bericht
news:1162218380.7...@b28g2000cwb.googlegroups.com...
--
Jelte
William
> As has been mentioned, AIX for PS/2 1.3 lacks functionality
> needed to make a build of Firefox work on it.
What functionality is missing AIX 1.3? I can only see the "GNU Syndrom":
To make Firefox work, you have to make gtk work that needs ..., that
needs ..., that needs ....
Regards,
Christian
Dillo builds for AIX 4.3.3+ would be interesting. As much as I love
Firefox and Gecko, it has grown to be a pretty heavy weight browser on
UNIX platforms.
> I think the real problem is the ancient version of GCC
> available on AIX 1.3.
What is it that made AIX 1.3 run so efficiently on our favorite
computers? Or is that a myth?
--
CL.
+-----------------------------------------+
| Charles Lasitter | Mailing / Shipping |
| 401/728-1987 | 14 Cooke St |
| cl+at+ncdm+dot+com | Pawtucket RI 02860 |
+-----------------------------------------+
The short answer is: "Attention to detail". The folks at Locus that
did the bulk of the work on AIX for the PS/2 knew that they would be
running the system on machines with limited processor power, limited
RAM and limited hard drive space, so it was important to optimize the
critical paths and keep all of the key system components small. In
truth, the same can be said of the other PC-focused Unix
implementations of the day, some did a better job of implementation
than others.
Very few of the AIX 1.2.1 or 1.3 machines I saw in IBM ran a GUI, as
GUIs suck up too many of those limited resources. As long as you stick
to the command line, AIX 1.3 runs pretty well, even on a 8580-311.
Rick Ekblaw
>> As has been mentioned, AIX for PS/2 1.3 lacks functionality
>> needed to make a build of Firefox work on it.
>
> What functionality is missing AIX 1.3?
Posix threads first. A working C++ Compiler, C++ libraries,
runtime-dynamic linking, a shell with enough resources for handling the
expanded command lines during the build and configure process...
On AIX 4, the GCC won't compile a working Firefox/Mozilla. You'll need
VisualAge Compiler 5.0.2 or later.
> I can only see the "GNU Syndrom":
> To make Firefox work, you have to make gtk work that needs ..., that
> needs ..., that needs ....
And the ability to allocate much more than 4MByte of continous RAM at once.
If anyone want's Firefox on a UNIX machine on PS/2 hardware, it would
suggest Solaris 2.6. But with only 64MB RAM max, Firefox won't work.
Perhaps it's possible with SCO UnixWare7...
If you want Firefox on PS/2 try with Windows NT4 or OS/2 Warp4.
If you want Firefox on Microchannel machines, get a RS/6000 and AIX
4.3.3 at least. I have given up to backport it to AIX 4.2.1, way too
much hassle.
Next time I promise stopping at three w's ;-)
"http://www.linkitup.de/ForAIX/Firefox"
"http://www.linkitup.de/ForAIX/Mozilla"
-
Uli
> Uli, what will work on 1.3? Firefox is well on it's weg to damnation, as
> far as I'm concerned. I have 1.5.0.7 and it's starting to slow down on
> my 95-Y with 128MB ECC under NT4 SP6a.
Lynx? NSCA Mosaic?
AIX 1.3 is a dead horse. Obsolete for more than a decade now.
You aren't faster riding 4 dead horses together.
Today many, many sites don't work anymore with Netscape 4 browsers.
Modern browsers are hungry monsters, eating up 50 to 100 megs of RAM
and hundreds of megahertz CPU is a bare minimum. It's not Firefox or
any other browser. It's today's web design with tons of Javascript,
stylesheets and animated graphics with transparency. Take a look at the
average code size of the average html page. All this code is to be
interpreted and rendered by the browser and host cpu.
You can use Firefox on a 7011-250, but it is slooooow. If there are
animated gif, jpeg or png pics on a page, the browser completely hogs
the CPU, close to deadlock the machine. I once did a test on a 7011-220
and the machine was constantly paging to death, even google needed a few
minutes to render. The browser GUI was an excellent exercise in patience
and hard disk testing.
A T4-P90 complex has roughly comparable performance to a 7011-250.
--
Uli
> Today many, many sites don't work anymore with Netscape 4 browsers.
> Modern browsers are hungry monsters, eating up 50 to 100 megs of RAM
> and hundreds of megahertz CPU is a bare minimum. It's not Firefox or
> any other browser. It's today's web design with tons of Javascript,
> stylesheets and animated graphics with transparency. Take a look at the
> average code size of the average html page. All this code is to be
> interpreted and rendered by the browser and host cpu.
>
> You can use Firefox on a 7011-250, but it is slooooow. If there are
> animated gif, jpeg or png pics on a page, the browser completely hogs
> the CPU, close to deadlock the machine. I once did a test on a 7011-220
> and the machine was constantly paging to death, even google needed a few
> minutes to render. The browser GUI was an excellent exercise in patience
> and hard disk testing.
> A T4-P90 complex has roughly comparable performance to a 7011-250.
> AIX 1.3 is a dead horse. Obsolete for more than a decade now.
> You aren't faster riding 4 dead horses together.
If it's dead, then let's render the bones for glue.
>> What functionality is missing AIX 1.3?
> Posix threads first.
Can't tell you about that, sorry. But as far I know, some Linux threads
are implemented as processes, so a fork() should do (except some details
;-)
> A working C++ Compiler, C++ libraries
Hmmm, GCC 2.95.3 and at least a stripped down gcc 3.0 (no fortran or
java) will work on 1.3.
> runtime-dynamic linking
Can't tell about this either. But if you mean the ".so stuff": why can't
it be compiled with static linking?
> a shell with enough resources for handling the expanded command
> lines during the build and configure process...
Bash 3.0 will run fine on AIX 1.3, most of the coreutils (except du and
maybe df) will work.
> And the ability to allocate much more than 4MByte of continous RAM at
once.
Is this a malloc() problem? I will check this theses days.
> If you want Firefox on PS/2 try with Windows NT4 or OS/2 Warp4.
Gee, my AIX 1.3 is running on a PII 400MHz box :-) I'm only having a few
troubles with the network ...
But anyway, I won't run Firefox on my AIX Box, this is a job for my
Windows "workhorse". I'm more interested in a "current" libc. As glibc
is a way too complex, I may take a look at newlib ore dietlibc.
Regards,
Christian
Some sort of web access via TCP/IP
Word 6
Some sort of graphical shell that has documentation, unlike TWM
>> A working C++ Compiler, C++ libraries
>
> Hmmm, GCC 2.95.3 and at least a stripped down gcc 3.0 (no fortran or
> java) will work on 1.3.
Have you run the GCC stress tests? Or only build the compiler itself?
The C++ compiler may build, but often fails to compile C++ code.
The C compiler is usually not a problem.
>> runtime-dynamic linking
>
> Can't tell about this either. But if you mean the ".so stuff": why can't
> it be compiled with static linking?
Some parts (gtk and glib) are cross-dependant. Very hard to static-link.
So dynamic linking is a mandatory feature. Runtime dynamic linking makes
thing a lot easier, as the symbols must not be resolved at binding the
objects. Another hurdle will be a working CORBA broker. This killed the
AIX 4.2.1 backport.
>> a shell with enough resources for handling the expanded command
>> lines during the build and configure process...
>
> Bash 3.0 will run fine on AIX 1.3, most of the coreutils (except du and
> maybe df) will work.
>
>> And the ability to allocate much more than 4MByte of continous RAM at
> once.
>
> Is this a malloc() problem? I will check this theses days.
>
When a (mostly) static build of Firefox is linked, the binder of AIX
4.3.3 allocates about 110 to 120 MByte of RAM. On HP-UX it is roughly
the same. The GCC/GNU binutils consume much more RAM than the vendor
tools on every platform I know. More than AIX 1.3 will handle.
The recursive make also consumes a few megs. I never was able to raise
the limits of AIX PS/2 more than factor 4 from the default, which will
be about 8 or 16MByte SHMMAX.
Those days 4 or 8 megs were a really huge amount of RAM for a single
process, when the biggest machines had 16 or 32 MByte max.
For e.g. UnixWare 1 and 2 came with a SHMMAX of 2 MByte default. To less
for bootstrapping a GCC 2.95 without raising the kernel limits.
I don't say, it's impossible. But I have about 100 downloads/users each
release. And the very most users are on AIX 5.2/5.3 in the meantime.
I won't pay months of work for something that may be impossible in the
end and if it works only three people in the world will try just for fun
before those three men realize it is completely useless in todays real
world.
--
Uli
ftp, nntp, email is possible
> Word 6
You can run Word 5.5 in Merge's dos box or better learn "vi".
> Some sort of graphical shell that has documentation, unlike TWM
what about mwm? It's working and well documented. It's configuration is
platform independant.
--
Uli
All of them threads are processes in Linux and that is the weak point. You
can just go and kill a single thread, convenient it might be, it is still
unnatural and risky.
> I don't say, it's impossible. But I have about 100 downloads/users each
> release. And the very most users are on AIX 5.2/5.3 in the meantime.
> I won't pay months of work for something that may be impossible in the
> end and if it works only three people in the world will try just for fun
> before those three men realize it is completely useless in todays real
> world.
I use your Firefox builds on my 7006-42T and 7012-397 - AIX 4.3.3 and
5.1. Many thanks, the AIX Netscape builds are excruciatingly slow.
If I could get xlC or VisualAge, I wouldn't need to static link all my
other software builds :-).
> If I could get xlC or VisualAge, I wouldn't need to static link all my
> other software builds :-).
GCC on AIX imposes many restrictions if you want to run the results on
any other machine, than the one it was compiled on.
Reply offline and I may help you improving your build environment a little.
--
Uli
> Firefox 2.0 has grown and is quite a lot slower than 1.5.0.7
So what did we get for this extra bloat?
Just extra phishing protection?
I still don't remember the compelling reason to upgrade.
I like Firefox because of the extensions, but for rendering and
speed I think Opera 8.54 still beats it.
My favorite feature is the session save and restore. I can now close
the browser and the sites, scrolling and text input will all come back
when I restart the browser. Inline spell checking is also helpful on
web forums and such. I've heard quite a few memory leaks were fixed,
and it has so far been more stable on my Linux machine (which used to
feel like a second-rate platform for build quality).
The big guns will come this May with Firefox 3 including a new major
version of the Gecko rendering engine. This release date sounds a month
or so ambitious to me however.
> My favorite feature is the session save and restore. I can now
> close the browser and the sites, scrolling and text input will
> all come back when I restart the browser.
The "Tab Mix PlusThis" extension offers this feature in a highly
configurable form for Firefox. It has been around for so long
that I just consider it a part of FireFox.
> Inline spell checking is also helpful on web forums and such.
I use the GNU Aspell for this, which integrates as an extension
for FireFox.
> I've heard quite a few memory leaks were fixed, and it has so
> far been more stable on my Linux machine (which used to feel
> like a second-rate platform for build quality).
1.0.5.7 is working pretty well for me.
> The big guns will come this May with Firefox 3 including a new major
> version of the Gecko rendering engine.
I hope they spend as much time as possible optimizing for speed.
> 1.0.5.7 is working pretty well for me.
1.5.0.8 is to be released this week. Security fixes.
My builds are from CVS and there were no checkins to the branch 1.5.0.8rc2.
>> The big guns will come this May with Firefox 3 including a new major
>> version of the Gecko rendering engine.
>
> I hope they spend as much time as possible optimizing for speed.
>
1.5 was faster than 1.0. So it *may* be possible.
But from what I've seen, the footprint will raise.
The very most users ask for features. Much more than legacy platforms.
--
Uli
> The very most users ask for features. Much more than legacy
> platforms.
I think it's unfortunate that browser designers feel pressured to
have things "built-in" for fear of being beaten over the head
with it otherwise.
So IE7 comes out and M$ says: "We have RSS, but FireFox
doesn't."
This is of course a crock, since there were many RSS extensions
available for FireFox long before IE7 came along.
I like the extension / plugin model that lets people add what
they want, and choose which implementation works the best for
them. Otherwise you're stuck with whatever comes as part of the
browser.