Hi Gábor,

On Mon, 5 Aug 2019, SZEDER Gábor wrote:

> On Sun, Aug 04, 2019 at 09:14:46PM +0200, Johannes Schindelin wrote:
> >
> > On Sat, 3 Aug 2019, SZEDER Gábor wrote:
> >
> > > So the dummy test above would start like this:
> > >
> > >   expecting success of 'commit works':
> > >           echo content >file &&
> > >   [...]
> >
> > Maybe it would make sense to also mention the test and test case number,
> > like so?
> >
> > 	expecting success of t9876.54 'it works':
>
> It's easy enough to do so, but I don't readily see any benefits.
>
> The '--verbose-log' of each test script is written to a separate
> file, whose name already contains the test number, so there is no use
> including it for each test case in there.  When running a test script
> and looking at its '--verbose' output, then surely all test numbers
> must be from that particular test script, so there is no use, either.
>
> As for the test case number, since the test cases are not numbered in
> the test scripts, I just ignore them right away, and look for test
> names anyway.
>
> Could you give an example?

Oh, my common way to read test logs is to sift through CI builds' logs,
in particular for the branches on https://github.com/gitster/git that
are based on commits without support for the convenient Tests tab in
Azure Pipelines.

So I hit Ctrl+F and look for `not ok` in literally hundreds/thousands of
lines. Of course, this is already the unrolled log _only_ of the failed
test scripts. Still, it is no fun to scoll back from test case 64 all
the way to 1 just to find out which test script contains this particular
failed test case.

Ciao,
Dscho