Axel Beckert writes ("Re: Bug#1000138: /usr/bin/screen: command arguments
beyond 62 silently discarded"):
> I thought about that a while (but didn't come to a conclusion quickly,
> so I only started a few weeks after the bug report), but while I agree
> that this is "data loss", the actual impact of this issue seems so
> small that it IMHO does not deserve RC status.
Fair enough.
> I'm also not really happy with the attached patch (from your second
> mail) and still undecided if I should really apply a patch that
> deliberately causes a crash.
>
> Then again, it might just be poorly named as it does not seem to cause
> a crash (like a segmentation fault), just exiting in panic.
Yes, by "crash" I didn't mean segfault.
> I'm removing the patch tag as the patch doesn't fix the issue but just
> makes it more visible.
Could you apply the patch in the meantime, though ? It is definitely
an improvement.
FYI, this bug really happened to a friend of mine, who was doing sme
large numerical processing tasks. They were constructing the screen
command line with a shell glob. The result was that some of their job
mysteriously simply didn't occur. This was confusing and my friend
spent some time poring over their Makefiles etc.
In another situation, the user might not have noticed, and might have
proceeded with outputs that were only calculated from some of the
inputs.
I think bailing completely is nearly always better than silently
forgetting half the job. This is why I felt this bug is RC: it can
cause data loss, and came quite close for my friend. With my patch it
is simply a bug, and one which the user can then work around.
Thanks,
Ian.
--
Ian Jackson <ijack...@chiark.greenend.org.uk> These opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
