Re: New ARI warning Sun Mar 4 01:53:53 UTC 2012

[Joel Brobecker <brobecker@adacore.com>]

Hi Yao,

I wish I had taken the time to answer a little sooner. Sorry about that.
I hope it doesn't give the impression that we're disinterested.

First of all, a big thank you for keeping on eye on the new ARI
regressions, it's always appreciated when one doesn't have to ping
about it.

> >> gdb/remote.c:10757: code: sprintf: Do not use sprintf, instead use xsnprintf or xstrprintf
> > gdb/remote.c:10757:      sprintf (rs->buf, ', use);
> 
> Looks like sprintf has been widely used in gdb/remote.c, so I'll
> give a separate fix.

That would be awesome. This behavior has been expanding just because
we're repeating the current patterns.

> >> gdb/common/agent.c:193: regression: strerror: Do not use strerror(), instead use safe_strerror()
> > gdb/common/agent.c:193:	       path, strerror (errno));
> 
> We don't have a safe_strerror in GDBserver, so I am inclined to wrap
> strerror by safe_strerror in GDBserver.

Not sure I understand what you are proposing here... Pedro is probably
your best reviewer anyway.

I had a look at safe_strerror, just to see what it's about. It's mostly
about not returning NULL, it seems. But the implementation is dependent
on the host (Windows or POSIX), so it could be a pain to move to common.

-- 
Joel