This is the mail archive of the
mailing list for the glibc project.
Re: [PATCH] stdlib/test-bz22786: Avoid spurious test failures using alias mappings
27.10.2018 11:44 Florian Weimer <firstname.lastname@example.org> wrote:
> * Rafal Luzynski:
> > [...]
> > The macro is actually named TIMEOUT rather than TIMEOUTFACTOR.
> > Maybe you meant "TIMEOUT factor" or "TIMEOUT macro" or "TIMEOUT value"?
> TIMEOUTFACTOR is an environment variable which can be used to scale
> TIMEOUT, for particularly slow systems.
OK, I didn't know about it. Therefore your comment is correct.
> >> [...]
> >> 1.72user 5.68system 2:05.25elapsed 5%CPU (0avgtext+0avgdata
> >> 920932maxresident)k
> >> 8453744inputs+0outputs (131431major+1145697minor)pagefaults 0swaps
> > Can you please wrap the log comment at 76 column or below?
> It's the standard output of the “time” command (not the shell built-in).
Yes, I understand this is an output of the command.
> > Did you mean this part to go to the commit comment?
> I just wanted to provide some justification for the complexity. I can
> drop it if it's unnecessary.
I agree this information is useful, I just think that the direct copy of
the output looks bad in the git log because the line is too long, also
the log command adds 4 more spaces at left which makes the line overflow
if it is wrapped at 80. I suggest you to write in human language something
like "execution time of the test goes from about 2 minutes down to about
1 second". I don't think it is necessary to insert detailed result like
"2:05.25", these results may and will vary. Also, I'm not even sure if
I understand correctly if "2:05.25" means "2 min 5.25 s" or anything
The core issue is that the test times out on some machines.
I didn't review carefully enough the rest of the patch, I can only confirm
that it works good and resolves the problem for me.