Another strange bug suggestion: inserting in wrong buffer when some screen related resources are scare (w32)

Another strange bug suggestion: inserting in wrong buffer when some screen related resources are scare (w32)

[Lennart Borgman]

This is another of these strange bug reports I usually sends. I have
not done so for a while since they are usually not used. However here
is a reminder of an old problem.

I just saw that Emacs was inserting text in the wrong buffer. This
seems related to running out of GDI objects.

I can't debug this. It is impossible since the state of Emacs or my pc
seems rather unreliable at this moment. However I think that maybe
Emacs can do something about it.

I have suggested before that Emacs should always check return values
from system calls. I am suggesting it again since it might catch a
problem like this. (And again: I would add more of them to my patched
version of Emacs if I had time to keep the patches, but I don't have
that.)

Re: Another strange bug suggestion: inserting in wrong buffer when some screen related resources are scare (w32)

[Stefan Monnier]

> I just saw that Emacs was inserting text in the wrong buffer. This
> seems related to running out of GDI objects.

I guess it's marginally better than "Emacs doesn't work", but it's still
pretty useless as a bug description.

> I have suggested before that Emacs should always check return values
> from system calls. I am suggesting it again since it might catch a
> problem like this. (And again: I would add more of them to my patched
> version of Emacs if I had time to keep the patches, but I don't have
> that.)

If you send your patches in clean manageable patches, we might be able
to incorporate them.  Tho of course, if they're not high enough on
people's list of priorities, they may linger for a long time.  But since
you can commit them yourself, all you need is for someone to approve it,
which is easier: the key is to make the patch short enough that it can
be reviewed with no effort.


        Stefan