jch wrote:
> Joachim Schipper wrote:
>> jch wrote:
>>> CONCLUSION: The performance of OpenOffice 2.2.1 on OpenBSD 4.2 (to be
>>> released in the fall) is still unacceptably slow. Since i am not an
>>> OBSD developer, i cannot comment on the underlying reasons for the poor
>>> performance when compared with OO running under Linux emulation under
>>> OBSD 3.9, or the native OO package under FreeBSD 6.2.

>>
>> I agree on that count. You may want to read the rest of the message,
>> though (in particular, test the same version of OO.o under Linux
>> emulation on 3.9 and 4.1 or -current).

> _____
> Joachim,
>
> Thanks for your feedback. Test already completed. Performance of OO
> 2.0.2 Linux emulation version is the same under 3.9 and -current.


So the emulated version is actually faster than the native version? That
is odd. Or worse than odd. Do both versions have java support? (Or both
not?) I don't see why that should matter that much, but it's the only
variable that makes sense.

>> Marc is almost certainly correct if he says this is not a widespread
>> problem.

> _____
> M.E. made no such statement.


"And btw, I'm afraid that what you're seeing is mostly a local problem."
indicated this, to me. Perhaps I was wrong.

>> In other words, if you want to get this resolved, you'll have to do
>> some of the work yourself, at the very least providing a proper bug
>> report. (no, this newsgroup is not 'official', and everything you
>> have posted up 'til now pieced together is not a proper bug report)

> _____
> I have not posted a bug report, because i have no idea if my
> measurements of OO's performance is a bug or not. That was the whole
> point of my post; (a) has this "problem" been noticed by others, (b) is
> it a bug, or (c) are there gcc compiler settings involved that might
> optimise the executables for speed, or (d) what? I don't read C-code,
> and cannot comment in that respect.


I don't think this is widely known, although - as I pointed out - I
don't really use OpenOffice.org a couple of times per year.

So, a) no, b) probably yes, and c) likely not.

>> And be at least not actively offensive to the people who you ask to
>> work for you.

> _____
> Sorry, Joachim, but i do not follow you. I do not believe that i have
> asked anyone to work for me. I also don't see that i have been
> "actively" offensive to anyone. If i have, it was certainly not
> intentional.


Maybe not, but I think the 'pull OO.o' comment was somewhat over the
top.

> I simply reported some measurements of how long it takes for the new
> OBSD OO 2.X.X package under OBSD 4.1 and -current to load, import/load
> .doc and .ppt files. Then i asked if anyone else had observed
> excessively long timing measurements. A Mr. Dave apparently has had the
> same experience as reported in this thread. He resorts to antiword as a
> solution. I tested FreeBSD's 6.2 new native OO package. The timing
> measurements were the same as those for the Linux emulation version
> under OBSD 3.9. Then i drew the conclusion that there must be an issue
> with the OO package under OBSD 4.1 & -current, and suggested what might
> be a reasonable course of action for the OBSD Development Team to
> pursue. Mr. deRaadt is the only one who can call the shots on who does
> what. All i can do as a user and OBSD financial supporter is seek
> information from the users at large, and to arrive at a suitable
> solution for what is a real problem with OO on my wife's older computer
> (450 MHz CPU, 512 Mb RAM). The answer was simple. I cloned my own OBSD
> 3.9 workstation, including OO 2.0.2 (Linux emulation version). It runs
> just fine, and OBSD 3.9 will stay on that machine as well as my own for
> a while yet.


Why? 4.1's emaulated OO.o is as fast as 3.9's, so you might as well
upgrade to get the other goodies (security updates, at least).

> To close, i want to mention that Mr. J. Schipper has always responded in
> a fair and rational manner to questions from users at large. I don't
> know if Mr. J. Schipper is a "certified member" of the OBSD Development
> Team, but if he is, he is a an example of a good ambassador for the
> OpenBSD "brand" of software.


No, the OpenBSD team consists of people who can actually write code, and
I don't belong in that category. Although I try to learn.

Unless my Java guess above happens to be correct, you may want to take
this to misc@openbsd.org - it attracts a much larger audience.

Joachim