This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [MI] Duplicate --thread-group flag not detected


> 2010-11-25  Marc Khouzam  <marc.khouzam@ericsson.com>
> 
> 	* mi/mi-parse.c (mi_parse): Missing else.

Normally, the man to approve this is Vladimir, so I can't approve, but
I had a quick look anyway.

I think that the patch above is correct, given the current implementation.
Another very bad scenario:

        <mi-command> --thread-group i1 --all --frame 1 --thread 1

I think that all the options after --thread-group will be ignored.

Speaking of the current implementation:

The whole parsing loop looked very odd at first, but I think I'm starting
to understand how it works, now.  I think it does work right now because
there is only 1 option that takes no argument ("--all").  So the code
'swallows' it first if found, and then checks for the other possible
switches.  It apparently tries to avoid error-handling duplication by
putting it at the end of the looping block, but then that leads to:

      if (*chp != '\0' && !isspace (*chp))
        error (_("Invalid value for the '%s' option"),
               start[2] == 't' ? "--thread" : "--frame");

... which is missing one case: --thread-group. As the number of options
being handled grows (for instance, I have in my TODO an entry for adding
support for "--task"), I think that we're going to have to rewrite this
slightly differently to make it easier to read and extend...

Another remark: It looks like the use of the --all/--thread-group/--thread
options with the -mi-break command is not documented? I could only see
them documented with commands such as -exec-continue....

Last comment: A small test would be nice...

-- 
Joel


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]