Re: [PATCH] test: Tnnn-* script file name renames

Subject: Re: [PATCH] test: Tnnn-* script file name renames

Date: Tue, 07 Jun 2016 17:28:14 +0300

To: David Bremner,


From: Tomi Ollila

On Tue, Jun 07 2016, David Bremner <> wrote:

> Tomi Ollila <> writes:
>> Over time, new test scripts of any particular "group" are added
>> in-between of current test scripts by giving them number that is not
>> divisible by 10. Eventually the space between groups fill up (i.e.
>> all 10 "slots" are used), or it is just hard to recognize suitable
>> number for new test.
>> Instead of trying hard planning work for grouping these and adding
>> more space for the groups (manually by hand) or trying to widen the
>> space for all, this (automatic) renumbering of the file name takes
>> care of the "problem" for a while.
>> ---
>> If this patch is applied, this will make all patches in queue that
>> have test script changes fail to merge :( -- so this is partially
>> wip, but at least commit message is written...
> I just rebased the retry-lock and message-properties series (separately)
> on top of your commit without conflicts, so it's probably not that big
> of a deal.
A-ha. it is nice that this works (it somehow notices renames... or it used
the source blob hashes to notice what to change)

> Do you think it's worth documenting what the supposed groups
> of tests are (maybe in tests/README)? The groups could really use some
> cleaning up. For example, we have at least two set groups of emacs
> tests. Some things will be harder to classify, of course. That's why we [1]
> believe in tags not folders ;). Nonetheless, perhaps it will be useful
> to have things a bit less chaotic.  Eventually we might even want to
> support the idea of running a "group" of tests.

As a quick thought is is just Simple Matter Of Documenting... I.e. how
to categorize tests to groups (what belong where when there are many
choices where files could go). But if someone(tm) can make such easily
understantable groups (and devices reasonable numbering -- i can adjust
my latest script (which does git-am'able patch files directly) to support
renumbering (if ever needed) (*) in this case too) it might be worth
sending for review...


(*) for reference: everything we tried there was still need to renumber C64
basic programs ;)
> d
> [1]: except those who believe in searchs, not tags.
> _______________________________________________
> notmuch mailing list