This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [PATCH] xcoffread.c: remove include of partial-stab.h
- To: Kevin Buettner <kevinb at cygnus dot com>
- Subject: Re: [PATCH] xcoffread.c: remove include of partial-stab.h
- From: Elena Zannoni <ezannoni at cygnus dot com>
- Date: Mon, 1 Oct 2001 17:36:29 -0400
- Cc: Elena Zannoni <ezannoni at cygnus dot com>, gdb-patches at sources dot redhat dot com
- References: <15287.35303.583879.469917@krustylu.cygnus.com><1011001210800.ZM19181@ocotillo.lan>
Kevin Buettner writes:
> On Sep 30, 5:08pm, Elena Zannoni wrote:
>
> > Turns out that xcoffread.c used only 2 of the many cases handled by
> > the switch in partial-stab.h. In order to reuse the whole file, a lot
> > of macros had to be defined to do nothing.
> >
> > I tested this patch on an aix4.3 system.
> >
> > I know this file has no official maintainer, but I'll wait a few days
> > to check it in anyway.
>
> I've looked this patch over. I've verified that only two cases of
> partial-stab.h were needed by xcoffread.c. Also, it appears to me
> that you've correctly integrated the two necessary cases into
> xcoffread.c.
Thanks for looking this over.
>
> My only concern about this patch is with the duplication of the two
> cases in xcoffread.c. These hunks of code are of substantial size and
> it occurs to me that correct maintenance of this code may require
> keeping the various copies in sync. I'm wondering if it'd be possible
> to turn these hunks of code into functions. That way we might be able
> to avoid duplicating (some of) this code in three different places.
>
Yes, that was the way I started thinking about this code too. But I
realized that it is probably better to completely decouple these
cases, so that a change to accomodate format A doesn't affect format B
or C. I think this common base was the reason for the 'fear' of
change in this code. By making these independent, it will become
easier to clean the code up w/o having to worry about other platforms,
that may be hard to test on (I had to dig around quite a bit to find
an AIX machine). I think the code can be simplified more this way
too, for instance we can get rid of the macro DBXREAD_ONLY, and a few
others.
> Kevin
Elena