[PATCH 0/6] Rebase of Pieter's "set test prereqs"

[Thomas Jost]

On Wed, 16 Nov 2011 10:53:42 -0800, Jameson Graef Rollins <jrollins at finestructure.net> wrote:
> On Wed, 16 Nov 2011 15:33:49 +0100, Thomas Jost <schnouki at schnouki.net> wrote:
> > Hello list,
> > 
> > This is another rebased version of Pieter's series to add GPG and Emacs as test
> > prereqs, plus some additions on my own. (Rebased and posted as requested by
> > Pieter [1].)
> > 
> > Changes as compared to Pieter's patches (including parts from [2]):
> > - prereqs are not tested using test_expect_success as they were in Pieter's
> >   original patches, but using a new function called test_set_bin_prereq. I wrote
> >   this before the gdb prereq was added, hence the different way to set it.
> 
> Hey, Thomas.  Thanks so much for this work.  This sounds like a better
> solution.
> 
> However, in the patches you send I see a lot of changes of the form
> 
>   -test_expect_success 'emacs delivery of encrypted message with attachment' \
>   +test_expect_success GPG 'emacs delivery of encrypted message with attachment' \
> 
> and
> 
>   -test_expect_equal \
>   +test_expect_equal GPG \
> 
> which seems to contradict what you've said above.  Not to mention that I
> don't see anything that modifies calls to the test_expect_ functions.
> Basically I see a lot more in the diffs than I would have expected in a
> cursory look.  Is this just a rebase flub, or is there something I'm
> missing?
> 
> jamie.

Hi Jamie,

I guess I wasn't clear in my explanations :)

Pieter's patches use this to detect the presence of GPG/Emacs and set
the prereq:

    +# GnuPG is a prereq.
    +test_expect_success "prereq: GnuPG is present" "which gpg" \
    +    && test_set_prereq GPG

There are 2 problems with this approach: 

- test_expect_success returns 0 regardless of the actual result of the
  command it runs. So even if gpg is not installed, 
    text_expect_success "..." "which gpg"
  will succeed, and "test_set_prereq GPG" will be run.
  This, however, has been fixed in commit 003e7180 -- which had not been
  pushed when I wrote this in the first place :)

- using test_expect_* to set a prereq does not make sense. If emacs is
  absent, the test suite would report a failed test. But a missing
  prereq is *not* a notmuch issue, so this should *not* be reported as a
  failed test.

Hence my first patch, which defines test_set_bin_prereq, a new helper
function to set a prereq without using any test_expect_*.

After that we can use the normal prereq syntax from the test suite:
- test_expect_success COMMAND --> run COMMAND, expecting it to succeed
- test_expect_success PREREQ COMMAND --> skip if PREREQ is not set, else
  run the test as before
(and same thing with the other test_expect_* functions)

Does it make more sense now?

Regards,

-- 
Thomas/Schnouki
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 489 bytes
Desc: not available
URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20111116/41cbc4ea/attachment.pgp>