develooper Front page | perl.perl5.porters | Postings from December 2000

Tests depending on extensions, and `make minitest'

Thread Next
From:
Stephen McCamant
Date:
December 31, 2000 15:36
Subject:
Tests depending on extensions, and `make minitest'
Message ID:
14927.44303.287182.823474@soda.csua.berkeley.edu
It appears that in the past there was an effort made to isolate the
tests that required extensions (not just the ones that tested
extensions) to the t/lib subdirectory. This made possible the
`minitest' make target, which uses miniperl to run all the tests
except those in t/lib, effectively testing the core without the
extensions. I for one found this target useful for testing changes to
the core, to avoid repeatedly rebuilding the ever-growing list of
extensions.

In recent development versions, however, make minitest always fails,
since several tests in other directories make use of extensions:

t/io/openpid.t uses FileHandle, which uses IO (via IO::File,
IO::Seekable and IO::Handle).

t/op/misc.t uses POSIX in a test of locales that was moved from
t/pragma/locale.t

t/op/my_stash.t uses Test, which uses Test::Harness, which uses
FileHandle, which uses IO as described above.

t/op/taint.t uses IPC::SysV to test the tainting of SysV-style IPC.

It was easy to fix openpid.t not use FileHandle (I have a patch if one
is wanted), but some of the others would require less trivial
changes. Is maintaining the isolation the right way to go, or should
we give up on it and come up with a more sophisticated way to tell
which tests minitest should run?

 -- Stephen McCamant


Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About