Re: [PATCH] Fix gdb crash due to SIGPIPE when the compile command fails

[Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>]

Hi Bernd,

> On 6/14/21 2:59 PM, Rainer Orth wrote:
>> Hi Bernd,
>> 
>>>> and I have a very hard time deciding where <signal.h> needs to be
>>>> included given the enormous range of headers included by compile.c.
>>>>
>>>
>>> Oh, I see, looks like it did only work by chance.
>>> Maybe just include it after all other headers:
>> 
>> that's what I've been using myself to fix the build locally.  However,
>> as I said I'm completely uncertain if that's the right fix, especially
>> given that gdb/compile files don't include any system headers directly.
>> 
>
> Are you concerned, that it might be unavailable on some platfroms?
>
> I seems to be included where needed in many gdb source files.
> It's not the plain system header anyways, actually it seems to be
> a gnulib-enhanced version which tries to improve the portability.

no, I'm concerned that including the system header directly doesn't
match gdb conventions.  Many are only included indirectly via some
gdb-private header, but I don't see a clear pattern there.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University