[petsc-dev] How test system options are handled

[Mark Adams]

Fair enough,

On Thu, Aug 7, 2025 at 3:09 PM Matthew Knepley <knepley at gmail.com> wrote:

> On Thu, Aug 7, 2025 at 2:30 PM Mark Adams <mfadams at lbl.gov> wrote:
>
>> It seems to me that if you want to override $args then you are changing
>> the test and can just change the code. Or it does not seem like a use case
>> that deserves to supported.
>>
>
> I use the test system all the time for development now, since it provides
> a convenient path for baseline options, incremental rebuilds, and
> verification. In this situation, overriding is very useful.
>
>    Matt
>
>
>> On Thu, Aug 7, 2025 at 6:24 AM Matthew Knepley <knepley at gmail.com> wrote:
>>
>>> We currently have two kinds of options in the test system. Options that
>>> come from the "args" line in the TEST section of the example source, which
>>> we will call $args, and options that come from the EXTRA_OPTIONS env var
>>> when the test is executed, which we will call $extra_args.
>>>
>>> Right now, in petsc_harness.sh:109, we feed the example executable the
>>> arguments
>>>
>>>   $extra_args $args
>>>
>>> This allows us to do things like declare EXTRA_OPTIONS="-fp_trap" for
>>> linux-pkgs-dbg, and then allow it to be turned off for certain tests with
>>> -fp_trap 0 in the argos line.
>>>
>>> This setup, however, does not allow us to override any options given in
>>> $args while running a test, which seems very restrictive. Should we create
>>> a third class of options?
>>>
>>>   Thanks,
>>>
>>>       Matt
>>>
>>> --
>>> What most experimenters take for granted before they begin their
>>> experiments is infinitely more interesting than any results to which their
>>> experiments lead.
>>> -- Norbert Wiener
>>>
>>> https://urldefense.us/v3/__https://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!b6oAz2fZQ4AfirMlFz4b4aayNowDlEBg5cwO2E3MHz6h1E5bGRCN0Gz6M0osVCe5bM8nRylS6c9G6NRweu2OOQ$ 
>>> <https://urldefense.us/v3/__http://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!es96ebvWwNhZgW9QAbhKl87uWMsmggCaWnv60yjXuLOqvK4co4KiyvhNTffA2KhyR42saIqyrZ93SWsl1pKX$>
>>>
>>
>
> --
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
> https://urldefense.us/v3/__https://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!b6oAz2fZQ4AfirMlFz4b4aayNowDlEBg5cwO2E3MHz6h1E5bGRCN0Gz6M0osVCe5bM8nRylS6c9G6NRweu2OOQ$ 
> <https://urldefense.us/v3/__http://www.cse.buffalo.edu/*knepley/__;fg!!G_uCfscf7eWS!b6oAz2fZQ4AfirMlFz4b4aayNowDlEBg5cwO2E3MHz6h1E5bGRCN0Gz6M0osVCe5bM8nRylS6c9G6NTh4OMfDg$ >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20250807/fe3e3c80/attachment-0001.html>