Re: Valgrind testing, ideas

[Sebastian Gniazdowski <psprint@zdharma.org>]

On 26 maja 2017 at 07:50:59, Bart Schaefer (schaefer@brasslantern.com) wrote:
> On Thu, May 25, 2017 at 10:04 PM, Sebastian Gniazdowski
> wrote:
> >
> > The arrlen branch seems to faint.
> 
> Sorry, what's the "arrlen branch"?

Mikachu's badarrays branch

> Not all tests use the single zsh. Some explicitly re-invoke
> $ZTST_exe, others run in subshells. In fact when I want to add
> valgrind to a "make check", I insert
> local ZTST_exe=(valgrind $ZTST_exe)

Thanks

> > Last thing, there should be much more test files. Single test per-file, I think.
> 
> I'm not thrilled with this suggestion. There are a lot of tests that
> depend on sharing a single %prep (and corresponding %cleanup) section,
> for example, and also a number of tests that depend on cascading one
> after the other. The amount of duplicated code that would result from
> attempting to make these into singleton files would lead to other
> problems.

Yes it is problematic, I've blindly created 28 test files from redis-adapted-V11db_gdbm, and noted that they fail, because database contents is cascaded through test, I rely on some keys from previous test to exist, etc. So I run all-tests in single file too, through valgrind, currently. But if lessening of load for eyes is wanted, more automatic-pin-pointing-capable output is wanted, then single test case per file is rather the way to go.

--
Sebastian Gniazdowski
psprint /at/ zdharma.org