On Tue, Dec 28, 2010 at 04:48:14PM +0100, Daniel Stenberg wrote:
> On Tue, 28 Dec 2010, Graeme Burton wrote:
> >I want to test the library thoroughly to make sure it's working,
> >and running ad hoc tests isn't really going to do an acceptable
> >job. I can see that there are Perl scripts which perform unit
> >tests; however I suspect that I won't be able to get hold of a Perl
> >interpreter for my mobile phone, much less a command line to run it
> >against. I propose to produce a C version of the Perl script and
> >use that for testing; however, as I'm new to libcurl development,
> >I'm very happy to accept all suggestions as to whether this is the
> >correct approach, and if it isn't, what approach I should take.
>
> This is basically the same problem that Pierre posted about the other
> day: the curl test suite is not properly adjusted to run and work
> fine on Windows.
>
> Also, I think you should consider doing the tests host-side and not
> on your target if you can't get a perl for it. Re-implementing the
> tests using something else seems like a lot of work for very little
> value.

The test suite also has hooks to allow "remote" testing on a platform
without perl support. I used it once to do limited testing on a Symbian
device. The requirements of the target device is a way to start curl with
arbitrary arguments (e.g. ssh) and a way to put and get files to/from the
device before/after the test (e.g. sftp or smb). The test framework, perl
scripts and test servers all run on the capable host machine. Some tests
won't work in this scenario, but the majority should work fine.

>>> Dan
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
Received on 2010-12-28