1. eu 4.0
- Posted by irv mullins <irvm at el??jay.com>
May 24, 2008
-
Last edited May 25, 2008
I downloaded yesterday's release.
Good news is it runs EuGTK programs.
Bad news it it takes 8+ seconds to start up.
Eu 3.1 takes 1 second or less.
Profile_time doesn't work in Linux, so where
or how should I start looking for the problem?
2. Re: eu 4.0
- Posted by Matt Lewis <matthewwalkerlewis at gmail.c??>
May 24, 2008
-
Last edited May 25, 2008
irv mullins wrote:
>
>
> I downloaded yesterday's release.
> Good news is it runs EuGTK programs.
> Bad news it it takes 8+ seconds to start up.
> Eu 3.1 takes 1 second or less.
>
> Profile_time doesn't work in Linux, so where
> or how should I start looking for the problem?
Did someone post a binary? I've run wxEuphoria apps with 4.0, and it
doesn't seem to take any longer.
Matt
3. Re: eu 4.0
- Posted by Jeremy Cowgar <jeremy at cowg??.com>
May 24, 2008
-
Last edited May 25, 2008
Matt Lewis wrote:
>
> irv mullins wrote:
> >
> >
> > I downloaded yesterday's release.
> > Good news is it runs EuGTK programs.
> > Bad news it it takes 8+ seconds to start up.
> > Eu 3.1 takes 1 second or less.
> >
> > Profile_time doesn't work in Linux, so where
> > or how should I start looking for the problem?
>
> Did someone post a binary? I've run wxEuphoria apps with 4.0, and it
> doesn't seem to take any longer.
>
In IRC, I gave rywilly a link to a built version because we are jointly working
on a wiki program. But I didn't make any public link. I'm sure others could have
downloaded it, but it's pre-pre-alpha. However, that being said, I ran a simple
bench mark of the sieve8k.exw script that comes with 3.1 and 4.0. To show nothing
has been altered, I ran the 3.1 version. Here are the results for 4.0 and 3.1 as
downloaded from RDS:
$ time /opt/euphoria-3.1/bin/exu /opt/euphoria-3.1/demo/bench/sieve8k.exw 10000
Prime Sieve Benchmark
Count: 1028
time: 4.82
real 0m4.825s
user 0m4.816s
sys 0m0.010s
$ time /opt/euphoria-4.0/bin/exu /opt/euphoria-3.1/demo/bench/sieve8k.exw 10000
Prime Sieve Benchmark
Count: 1028
time: 2.75
real 0m2.750s
user 0m2.746s
sys 0m0.003s
Notice that 4.0 is more than 2 seconds faster. Now, this is including start
up/parsing/run/cleanup time as the unix time command starts the timer, then
spawns the program it's timing, then only stops the timer one the spawned program
has been terminated and returned control to it.
Why is it taking 8 seconds? That seems very odd. Can you tell us where you got
the binary or if it is compiled from sources? If compiled from sources, what
revision of svn?
Thanks.
--
Jeremy Cowgar
http://jeremy.cowgar.com
4. Re: eu 4.0
- Posted by Jeremy Cowgar <jeremy at cowg?r.c?m>
May 24, 2008
-
Last edited May 25, 2008
To further test simply startup speed I made this bash script:
#!/bin/sh
for i in `seq 1 10000`;
do
/opt/euphoria-3.1/bin/exu hello.e
done
Each version of course calls the correct version.
hello.e is:
integer a
a = 1
$ time ./test-3.1.sh
real 0m11.550s
user 0m4.153s
sys 0m7.193s
$ time ./test-4.0.sh
real 0m11.288s
user 0m3.786s
sys 0m6.476s
So the above loads/parses/executes/terminates exu 10,000 times. 4.0 is a tad
faster, not much, but a tad.
Something must be terribly wrong with the 8 second startup time for 4.0 and 3.1
being less than a second. Is this repeatable? Is there a test script we can run
to try and debug the problem you are experiencing?
--
Jeremy Cowgar
http://jeremy.cowgar.com
5. Re: eu 4.0
- Posted by Bernie Ryan <xotron at blue?rog.c?m>
May 24, 2008
-
Last edited May 25, 2008
Jeremy Cowgar wrote:
>
> To further test simply startup speed I made this bash script:
>
> #!/bin/sh
>
> for i in `seq 1 10000`;
> do
> /opt/euphoria-3.1/bin/exu hello.e
> done
>
> Each version of course calls the correct version.
>
> hello.e is:
>
> }}}
<eucode>
> integer a
> a = 1
> </eucode>
{{{
>
> $ time ./test-3.1.sh
>
> real 0m11.550s
> user 0m4.153s
> sys 0m7.193s
>
> $ time ./test-4.0.sh
>
> real 0m11.288s
> user 0m3.786s
> sys 0m6.476s
>
> So the above loads/parses/executes/terminates exu 10,000 times. 4.0 is a tad
> faster, not much, but a tad.
>
> Something must be terribly wrong with the 8 second startup time for 4.0 and
> 3.1 being less than a second. Is this repeatable? Is there a test script we
> can run to try and debug the problem you are experiencing?
>
> --
Jeremy:
When Rob came out with version 2.5 the user's had a similar problem with
EXW.
http://www.openeuphoria.org/cgi-bin/esearch.exu?thread=1&fromMonth=B&fromYear=9&toMonth=1&toYear=A&keywords=%22v2.5+Opens+exw+files+wayyyy+too+slow%22
Bernie
My files in archive:
WMOTOR, XMOTOR, W32ENGIN, MIXEDLIB, EU_ENGIN, WIN32ERU, WIN32API
Can be downloaded here:
http://www.rapideuphoria.com/cgi-bin/asearch.exu?dos=on&win=on&lnx=on&gen=on&keywords=bernie+ryan
6. Re: eu 4.0
- Posted by Matt Lewis <matthewwalkerlewis at gm?il.com>
May 24, 2008
-
Last edited May 25, 2008
Bernie Ryan wrote:
>
> When Rob came out with version 2.5 the user's had a similar problem with
> EXW.
That's true, but it was due to the switch to a translated front end from
hand coded C. Most of that was fixed. There's much less difference between
3.1 and 4.0.
Matt
7. Re: eu 4.0
Jeremy Cowgar wrote:
>
> Why is it taking 8 seconds? That seems very odd. Can you tell us where you got
> the binary or if it is compiled from sources? If compiled from sources, what
> revision of svn?
>
> Thanks.
>
> --
> Jeremy Cowgar
I got it from this weird guy's website. Maybe you've heard of him :)
http://jeremy.cowgar.com/files/eu40.tar.gz 23-May-2008 17:20 3.1M
8. Re: eu 4.0
irv mullins wrote:
>
>
> I got it from this weird guy's website. Maybe you've heard of him :)
>
> <a
> href="http://jeremy.cowgar.com/files/eu40.tar.gz">http://jeremy.cowgar.com/files/eu40.tar.gz</a> 23-May-2008
> 17:20 3.1M
You downloaded from him? Wow, now we all know your crazy! 
Anyway, that's the same binary (well, w/in a few revisions of) from the tests
that I showed here. Can you give us an example that's messing up on you?
--
Jeremy Cowgar
http://jeremy.cowgar.com
9. Re: eu 4.0
- Posted by irv mullins <irvm at ellij?y?com>
May 28, 2008
-
Last edited May 29, 2008
Jeremy Cowgar wrote:
>
> irv mullins wrote:
> >
> >
> > I got it from this weird guy's website. Maybe you've heard of him :)
> >
> > <a
> > href="http://jeremy.cowgar.com/files/eu40.tar.gz">http://jeremy.cowgar.com/files/eu40.tar.gz</a>
> 23-May-2008 17:20 3.1M</font></i>
>
> You downloaded from him? Wow, now we all know your crazy!
>
> Anyway, that's the same binary (well, w/in a few revisions of) from the tests
> that I showed here. Can you give us an example that's messing up on you?
>
Yes, actually about 11,000 warnings were being generated. Something
to do with the difference in the way includes are handled between 3.1 and 4.0.
I've fixed those in my eugtk code.
There are still some 'interesting' developments. More on those later.
10. Re: eu 4.0
- Posted by Matt Lewis <matthewwalkerlewis at g?ai?.com>
May 28, 2008
-
Last edited May 29, 2008
irv mullins wrote:
>
> Jeremy Cowgar wrote:
> >
> > Anyway, that's the same binary (well, w/in a few revisions of) from the
> > tests
> > that I showed here. Can you give us an example that's messing up on you?
> >
> Yes, actually about 11,000 warnings were being generated. Something
> to do with the difference in the way includes are handled between 3.1 and 4.0.
That would explain the slow down. The issue is that you were using something
from a file that wasn't directly or indirectly included from the file using
the symbols.
Matt
11. Re: eu 4.0
- Posted by Jeremy Cowgar <jeremy at c?wg?r.com>
May 28, 2008
-
Last edited May 29, 2008
irv mullins wrote:
>
> Yes, actually about 11,000 warnings were being generated. Something
> to do with the difference in the way includes are handled between 3.1 and 4.0.
>
> I've fixed those in my eugtk code.
>
Did that fix anything in regards to speed?
> There are still some 'interesting' developments. More on those later.
I hate to say this, but it's good you are doing this. It is giving us an
opportunity to see what a large project is going to do with 4.0. Euphoria has a
new wiki, it's not official yet, but I wonder if you would be so kind to add to a
wiki page I've created:
http://openeuphoria.org/wiki/euwiki.cgi?MigratingTo40
The reason it is not official yet is the parsing engine is a bit sensitive to
correct syntax. You can goof it up if you try, which should not be so. Soon I
will announce its official use, but while things are fresh in your mind, please
document the changes there if you do not mind.
That goes for everyone else as well. Thanks!
--
Jeremy Cowgar
http://jeremy.cowgar.com
12. Re: eu 4.0
- Posted by ChrisBurch3 <crylex at g?ai?.com>
May 28, 2008
-
Last edited May 29, 2008
Euphoria has a wiki? Wow. Nice.
Any chance of a forum? (Or arranging threads on the list by date last posted)?
Chris
13. Re: eu 4.0
- Posted by Jeremy Cowgar <jeremy at co?g?r.com>
May 28, 2008
-
Last edited May 29, 2008
ChrisBurch3 wrote:
>
> Euphoria has a wiki? Wow. Nice.
>
> Any chance of a forum? (Or arranging threads on the list by date last posted)?
>
Sure: http://www.openeuphoria.org/EUforum/threads.html
--
Jeremy Cowgar
http://jeremy.cowgar.com
14. Re: eu 4.0
Jeremy Cowgar wrote:
>
> ChrisBurch3 wrote:
> >
> > Euphoria has a wiki? Wow. Nice.
> >
> > Any chance of a forum? (Or arranging threads on the list by date last
> > posted)?
> >
>
> Sure: <a
> href="http://www.openeuphoria.org/EUforum/threads.html">http://www.openeuphoria.org/EUforum/threads.html</a>
>
>
>
No, it doesn't so that (yet). Check out the date of this thread on the threads
page, check the date of the last post, and check its position in the list
of threads. Ideally (IMHO) the thread with the most recent posting should
be sorted to the bottom of the page - that would be really cool.
Chris
> --
> Jeremy Cowgar
> <a href="http://jeremy.cowgar.com">http://jeremy.cowgar.com</a>
15. Re: eu 4.0
ChrisBurch2 wrote:
>
>
> No, it doesn't so that (yet). Check out the date of this thread on the threads
> page, check the date of the last post, and check its position in the list
> of threads. Ideally (IMHO) the thread with the most recent posting should
> be sorted to the bottom of the page - that would be really cool.
>
Ah, Ok. got ya. I do not think we will be replacing EUforum.
--
Jeremy Cowgar
http://jeremy.cowgar.com
