[PATCH v3 0/3] gas: add command line option to inhibit info messages

[Jan Beulich]

On 26.07.2024 12:59, Matthieu Longo wrote:
> On 2024-07-10 13:27, Jan Beulich wrote:
>> On 10.07.2024 13:26, Matthieu Longo wrote:
>>> Hi,
>>>
>>> GAS currently emits information messages for context information along warnings. These extra messages might sometimes pollute the test expectations when checking for error message patterns in stderr output.
>>>
>>> This patch aims at providing two new flags while preserving the existing behavior if none of the options is provided.
>>>    * --info, similar to the existing --warn flag to enable diagnostic information messages (default behavior).
>>>    * --no-info, similar to the existing --no-warn flag to disable diagnostic information messages.
>>>
>>> It also adds the flags to the existing documentation, and command manual.
>>>
>>> As an example of its benefit, I used the new flag for the system register tests under gas/testsuite/gas/aarch64/sysreg testsuite, and the info message for macro expansion were removed inside the tests expectations.
>>>
>>> Regression tested on aarch64-none-elf, and no regression found.
>>>
>>> Ok for binutils-master?
>>>
>>> Regards,
>>> Matthieu.
>>>
>>> Changes from v0:
>>> - reorder warning options
>>>    https://sourceware.org/pipermail/binutils/2024-June/135232.html
>>> - rename flag_no_infos to flag_no_information.
>>> - documentation: move --no-info and --info flags after warnings flags.
>>> - remove from this patch's scope the tests relying on the line number.
>>>    add a check in output_info() in AArch64 backend.
>>>    check flag_no_information flag and quickly return if enabled, as requested in the previous revision's review.
>>>    https://sourceware.org/pipermail/binutils/2024-June/135231.html
>>>
>>> Changes from v1:
>>> - move the '--info' section up to respect alphabetical order.
>>
>> It's at least close to nitpicking, I know, but still: In response to
>> this change of yours I said that you didn't go far enough with what you
>> did. I did not ...
>>
>>> Changes from v2:
>>> - reorder '--no-info' sections to respect alphabetical order.
>>
>> ... ask for ordering to be done based on the "no-" prefix. The place
>> I'd look for when wanting to know something info message related would
>> be at 'i', not 'n'. I'm sorry if this didn't come through clear enough.
>>
>> The fact that --no-pad-sections lives where it does is, I guess, just
>> because there's no --pad-sections (for whatever reason).
>>
>> And again, commenting would have been easier if you sent patches inline,
>> as then there is context to actually associate a comment with, and it is
>> also possible to talk in "further up" and "further down" terms.
> 
> I merged the first patch of the series as you already approved it.
> I would like to send again the 2 remaining patches *INLINED* in the 
> email (as you requested) and UNCHANGED (=v3) so that it will be easier 
> for you to comment.
> I apologize for not doing this the first time, I am not familiar with 
> practices of patch reviews via emails.
> 
> Regarding the ordering of options, I decided to move --no-info and 
> --info flags to the letter 'N' to follow the same approach as with 
> --no-warn. In my understanding, '--no-warn' (a.k.a -W) is referenced at 
> the letter 'W' because of its shortened option '-W', and not because of 
> '--warn'.
> The current listing of the options in gas doc is a bit confusing, and 
> above all, having the '-W' short option for `--no-warn` instead of 
> `--warn` is utterly confusing from a user's perspective.
> 
> Additionally, I want to highlight that --warn and --info are enabled by 
> default, so the --no-* variant might be the first one we want to list.

Personally I didn't fully agree with this re-arrangement that you did there
also for --warn, but not enough to loudly object. Yet I continue to think
info-related entries want to sort at 'i', not at 'n', not matter that
there's no shorthand option.

Jan