[m-rev.] for post-commit review: update comments in file_names.m
Zoltan Somogyi
zoltan.somogyi at runbox.com
Fri Jul 7 19:23:17 AEST 2023
On 2023-07-07 10:36 +02:00 CEST, "Julien Fischer" <jfischer at opturion.com> wrote:
>> + ---> ext_obj_dollar_o % ".$O"
>> + ; ext_obj_dollar_efpo % ".$(EXT_FOR_PIC_OBJECTS)"
>> + ; ext_obj_o % ".o"
>> + ; ext_obj_pic_o % ".pic_o"
>> + ; ext_obj_obj_opt % object_file_extension option
>> + ; ext_obj_pic_obj_opt. % pic_object_file_extension option
>
> I suggest adding a comment that ".obj" is handled by the ext_obj_dollar_o case.
When generating Mmakefile fragments. I expect that it is handled
by the object_file_extension option when writing the file itself,
but I can't find where that is done. Do you know?
>> + ---> ext_mf_d % ".d"
>> + ; ext_mf_dv % ".dv"
>> + ; ext_mf_dep % ".dep"
>> + % XXX DODGY This extension is use in only one place, in write_deps_file.m.
>> + % It looks strange to me (zs), because I have no idea where ".dir"
>> + % comes from, or what system component puts object files there.
>> + ; ext_mf_dir_sl_all_os. % ".dir/*.$O"
>
> I don't recall even seeing that before.
Then you share my bafflement.
I suppose the only thing we can do is to delete
all references to .dir, including this one, and the
dirs_subdir mmake variable, and see whether
anything breaks :-(
>> + ; ext_misc_ngs_prof. % ".prof"
>> + % XXX DODGY Given that different profiling grades generate
>> + % different profiles (specifically, they produce different subsets
>> + % of the whole set of kinds of info that the non-deep profiler can
>> + % generate), shouldn't this be a grade-specific extension?
>
> I'm going to make a guess and say that was just overlooked. (Given that it's
> rare to install more than one .prof grade I can see that happening.)
Thanks for that. I want with all your suggestions.
Zoltan.
More information about the reviews
mailing list