This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFA 0/5] Make cp-name-parser.y a pure parser
- From: Pedro Alves <palves at redhat dot com>
- To: Joel Brobecker <brobecker at adacore dot com>, Tom Tromey <tom at tromey dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Fri, 1 Jun 2018 13:27:09 +0100
- Subject: Re: [RFA 0/5] Make cp-name-parser.y a pure parser
- References: <20180530142953.24934-1-tom@tromey.com> <20180531000248.2r457ch2gfsajjy4@adacore.com>
On 05/31/2018 01:02 AM, Joel Brobecker wrote:
> Hi Tom,
>
>> I've been experimenting a bit with using threads to parallelize
>> psymbol reading.
>>
>> One difficulty here is that cp-name-parser.y is used by the DWARF
>> reader, but is not reentrant.
>>
>> It turns out, though, that fixing this is not too difficult. I had at
>> first thought that this would require Bison, but Byacc also supports a
>> pure mode. More details in patch #3.
>>
>> Tested by the buildbot. I'm unaware of whether byacc is used by any
>> of the builders, so I also built gdb using byacc for this parser and
>> locally ran gdb.cp/*.exp as a smoke test.
>
> I've scanned the patch series, and it looked good to me. But I'm not
> a big expert on yacc files, so could you give it a couple of weeks
> for others to have a chance to review also? If no comments was sent
> by then, go ahead and push.
It looks fine to me too. I'm a little surprised to see a parser
state object pointer being passed around instead of making the
functions be methods of the parser type, but I guess it's really
the same either way.
Thanks,
Pedro Alves