Re: [PATCH v3] Make '{add-,}symbol-file' not care about the position of command line arguments

[Sergio Durigan Junior <sergiodj@redhat.com>]

On Thursday, November 30 2017, Pedro Alves wrote:

> On 11/30/2017 01:33 PM, Sergio Durigan Junior wrote:
>
>> +# Since we're here, might as well test the 'symbol-file' command and
>> +# if its arguments can also be passed at any position.
>> +gdb_test "symbol-file -readnow $binfile" \
>> +    "Reading symbols from ${binfile}\.\.\.expanding to full symbols\.\.\.done\." \
>> +    "symbol-file with -readnow first"
>> +clean_restart
>> +gdb_test "symbol-file $binfile -readnow" \
>> +    "Reading symbols from ${binfile}\.\.\.expanding to full symbols\.\.\.done\." \
>> +    "symbol-file with -readnow second"
>> +gdb_test "symbol-file -readnow" \
>> +    "no symbol file name was specified" \
>> +    "symbol-file without filename"
>> +gdb_test "symbol-file -- non-existent-file" \
>> +    "non-existent-file: No such file or directory\." \
>> +    "symbol-file with -- disables option processing"
>
> This should be "-non-existent-file", with leading "-" ...
>
> Missing the same test with "add-symbol-file".

Right, fixed.

> Another thing I wondered is why do we have the
> expecting_sec_name/expecting_sec_addr variables instead of
> simply reading the arguments in the argv ahead inline
> like maintenance_print_symbols does?
>
> As is, the patch essentially forbids section names
> that start with '-':
>
>  (gdb) add-symbol-file ./gdb 0 -s -funnysection 0
>  add symbol table from file "./gdb" at
>          .text_addr = 0x0
>          -funnysection_addr = 0x0
>  (y or n) 
>
> =>
>
>  (gdb) add-symbol-file ./gdb 0 -s -funnysection 0
>  Missing section address after "-s"
>
> and I'm not sure that's really a good idea.  I don't think
> we should really interpret/require section names to be in any
> way/format.

I agree; I hadn't thought of this case before.  I implemented the
argument reading ahead, as proposed.

> (Another thing that I noticed but I'm kind of ignoring is the fact
> that gdb_test treats the question as optional gdb output, so
> pedantically gdb could stop outputting the question and answer
> "n" automatically and the testcase wouldn't notice.)

Would you prefer if I made the tests answer "y" instead?  Or maybe I'm
misunderstanding your concern.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/