This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH v3 2/2] compile: Add 'set compile-gcc'
- From: Pedro Alves <palves at redhat dot com>
- To: Jan Kratochvil <jan dot kratochvil at redhat dot com>
- Cc: gdb-patches at sourceware dot org, Phil Muldoon <pmuldoon at redhat dot com>
- Date: Mon, 27 Apr 2015 20:24:16 +0100
- Subject: Re: [PATCH v3 2/2] compile: Add 'set compile-gcc'
- Authentication-results: sourceware.org; auth=none
- References: <20150423203402 dot 23140 dot 92757 dot stgit at host1 dot jankratochvil dot net> <20150423203413 dot 23140 dot 35224 dot stgit at host1 dot jankratochvil dot net> <20150423210815 dot GA8626 at host1 dot jankratochvil dot net> <553E5A1E dot 5000003 at redhat dot com> <20150427175445 dot GA12679 at host1 dot jankratochvil dot net>
On 04/27/2015 06:54 PM, Jan Kratochvil wrote:
> On Mon, 27 Apr 2015 17:47:42 +0200, Pedro Alves wrote:
>> IIUC, gdb will always apply the same search as when this is set
>> empty? That is, the user can also set this to a regex. So it seems to me
>> that the documentation (manual and help) doesn't match the implementation?
>
> That it can be also a regex is an API bug because I wanted to make a minimal
> API change. Rather than officially documenting such bug I find then better to
> rather make a proper complex change to the API. Given that you requested an
> API rework anyway I will try to post the new API even with this change.
This overload had given me lots of pause, and trying to think it through
(it wasn't clear what the intention was), it seemed to me that it kind of made at
least some sense to allow specifying a different regex, but details of the search
algorithms are foggy to me. It probably really doesn't make sense to overload.
Given we now clearly understand how to add new methods and it isn't that complex,
and we're already bumping the API, yes, let's please avoid an overload hack
when we don't need it, avoiding such confusions.
Thanks,
Pedro Alves