Re: [PATCH] Add completer for skip numbers

[Simon Marchi <simon.marchi@polymtl.ca>]

On 2018-11-10 14:07, Pedro Alves wrote:
> On 11/10/2018 05:39 PM, Simon Marchi wrote:
>> Add completer to various commands that accept skip numbers:
>> 
>>   - skip enable
>>   - skip disable
>>   - skip delete
>>   - info skip
>> 
>> These commands also accept ranges, but I am not too sure of how to do
>> that properly, so I went for the simpler goal of complete just 
>> numbers.
>> 
>> A future idea would be to make a re-usable and well-tested completer 
>> for
>> numbers and ranges.  I think it could at least be re-used for 
>> breakpoint
>> numbers (for example with the "enable breakpoints" command).
> 
> And threads.

Right.


> Please use the lib/completion-support.exp routines
> for testing this.  Those exercise both the complete command
> and actual TAB completion.  Above you want to use
> test_gdb_complete_multiple.
> 
> You should also add:
> 
> - a test that exercises unique completions like "skip delete 12",
>   with test_gdb_complete_unique.
> 
> - a test that exercises no completions at all, like "skip delete 2",
>   with test_gdb_complete_none.
> 
> - some test like "skip delete a1" to make sure we don't
>   mistakenly complete that a1 into a1/a10/a11/a12.
> 
> As for a range completer, doesn't e.g., "skip delete 1-2<tab>" already
> try to complete the "2", and "skip delete 1-" present all the numbers?
> I'd think so, given that '-' is part of the default word break chars
> set (default_word_break_characters).  You should add tests for that 
> too,
> IMO, since people will very naturally use it.  Even if not super smart
> (e.g. 2-[TAB] ideally wouldn't present "1"), it's still useful as is.

Thanks for all the feedback, I hope I have addressed everything in v2.

Simon