[docs] [issue11260] smtpd-as-a-script feature should be documented and should use argparse

Xavier Morel report at bugs.python.org
Wed Feb 23 08:21:24 CET 2011


Xavier Morel <xavier.morel at masklinn.net> added the comment:

> Only "document and formalize" the parts that are well thought out.

I don't believe I have the knowledge, right or ability to make that call for any module or package but a pair of extremely obvious ones (http.server seems a pretty good candidate as it's documented in the module doc, and is just missing a CLI help). Should I create a bug and nosy the people who have been involved in the module to get their opinion for each one? (as with this one, but maybe with a different wording/initial proposal) For this one, David expresses concerns on the stability on the interface (and the point of the idea), you express even bigger concerns with the idea itself and more generally modules-as-scripts in the stdlib (not their existence so much as their official support, which full documentation would imply, if I didn't misunderstand you), and Barry seems to okay the idea as long as an extensive test suite is created beforehand to ensure there is no regression, is that sufficient to go ahead with more work and defer the final decision for when that will be done?

Also, if this is to become an actual project if the smtpd stuff ever reaches fruition (though not necessarily a big or impactful one), should there be a meta-bug? (I know they're used in many bugzilla projects and I see the tracker handles dependencies)

> There's no harm in adding --help or a usage example, but please avoid making behavior guarantees that we really don't want to have to live with.

I'm not sure what that results in: for "quick hacks" which are not to be officially supported, does this mean fixing the CLI (adding a help for tools missing one) is OK but no more? Is an argparse switch still acceptable? Tests for the CLI tool? I'm guessing documentation in the module would definitely be off limits for those, is my interpretation correct? And again, who would be allowed to make the call on which modules-as-scripts may be considered supported, and can't be?

With this one, I created separate patches for the documentation and the CLI parsing alterations, which would allow merging the CLI without adding it to the official documentation (which would I think imply a lack of official support), would that be OK for future works, if this one pans out?

----------

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue11260>
_______________________________________


More information about the docs mailing list