Mail Archives: cygwin-developers/1998/08/09/21:29:45
On Sun, Aug 09, 1998 at 08:26:40PM +1000, Trevor Yann wrote:
>I was thinking that a test suite for the cygwin dll would be a good thing.
>The only specific test that I have seen is test.c in the winsup directory,
>and it only performs very rudimentary tests of the fork code.
>
>I would like to have tests using "straight" win32 code either mingw or msc.
>These could test the win32 environment provided by the various flavours of
>win 95/98/nt. These would help confirm whether work arounds are required in
>the cygwin dll. Consider the number of the versions of win 95 that are
>around - original, service pack 1, oem 2.1 and oem 2.5 (did I miss one?),
>each having a different set of bugs with the win32 api implementation.
>Likewise NT has a number of different versions that will each differ in
>their implementation of the win32 api.
>
>I am willing to start this sort of test suite for the win32 api, but it
>should be done within a testing framework. DejaGnu comes to mind, but it
>doesn't currently run under the cygwin environment.
>
>I have hacks to tcl to allow shell scripts to be run, but the required
>changes for expect are beyond me. The expect code seems to assume a unix
>environment, but the tcl that is used is very "windozsy".
>
>Does anyone have a version of dejagnu that works under the cygwin
>environment? A working dejagnu is necessary for me to start any test suite.
I have dejagnu working with my current .dll but it required changes to expect
and a minor change to dejagnu. This .dll will be released when I'm certain
that there aren't any egregious bugs left in it.
We (Cygnus) are also committed to performing more testing on the product.
--
cgf AT cygnus DOT com "Everything has a boolean value, if you stand
http://www.cygnus.com/ far enough away from it." -- Galena Alyson Canada
- Raw text -