On Wed, May 29, 2024 at 05:24:44PM +0800, Qian Yun wrote:
>
> On 5/27/24 22:10, Waldek Hebisch wrote:>
> > > > > What's its CPU model and how many threads are you using to build
> > > > > the book?
> > > >
> > > > Quad core Core 2 2.4.GHz. I just did 'make book', that is serial
> > > > build.
> > > >
> > >
> > > I limit my CPU to 800MHz, but still can't reproduce it.
> >
> > I suspect that this may disc drive. I have mirrored pair of
> > hard discs. After large writes processes writing to disc effectively
> > stop for few seconds, but CPU bound go forward.
> >
>
> Can't reproduce it on my mechanical hard drives either.
>
> From your description it looks like it's the missing lines
> at the end problem.
I did more trial. With diff:
diff --git a/src/sman/sman.c b/src/sman/sman.c
index 652e270..5b0c861 100644
--- a/src/sman/sman.c
+++ b/src/sman/sman.c
@@ -773,10 +773,10 @@ monitor_children(void)
}
switch(proc->death_action) {
case Die:
- fricas_sleep(100); /* wait a bit for socket buffer to be processed */
+ fricas_sleep(500); /* wait a bit for socket buffer to be processed */
kill_all_children();
clean_up_sockets();
- fricas_sleep(100);
+ fricas_sleep(500);
exit(0);
case NadaDelShitsky:
break;
first build went fine, but a few later failed. Actually, it looks
worse than previous version where probability of success looked
higher.
Yes, problem is because some .tex files are truncated. In one run it
was 'ug10.tex', in few cases it was 'SEGBIND.tex'. In other cases
I did not check the files but LaTeX error messuge indicated truncation.
> Do you think this is a problem that needs to be fixed before commit?
We should understand what is happening. As I wrote, HyperDoc and
graphics protocal is inherently racy, but in previously it worked
relatively well. Ocasional hangs did happen. But truncated files
is a new thing. AFAICS book build is similar to build of HyperDoc
pages, parts that differ do not involve FRICASsys, HyperDoc or graphics.
On this machine I have build HyperDoc pages thousends of times
and I believe that each time build was correct. There is no 'xvfb'
on this machine and building graphic pages pops images on the screen,
so I build graphic pages less frequently, but I think it was
hundreds of times. I only recently tried to build book on this
machine as book needs extra LaTeX packages. So I do not know if
degradation was already in previous book process or is a new thing.
But clearly there is something fishy with current build process
compared to build of HyperDoc pages.
--
Waldek Hebisch
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to fricas-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/fricas-devel/ZldA4PneuEZpLNji%40fricas.org.
