Re: Greg testing framework

[Matt Rice]

Hi,

>Hi,
>
>
>We are going to use Greg (gnustep-guile testing
>framework) for
>Hégémonie, I'm wondering how I should integrate it
>with the makefiles.

haven't thought about this, sounds like a good idea to
me though

>
>I have also only been able to test our own classes,
>exporting them in a
>library, installing this library on the system, and
>doing a
>(dynamic-link "myClasses.so") in the greg guile
>script. Is there a clean
>way to test the classes, without having to install
>them ?

i dunno your definition of clean, and which side of
that this would fall under but..
maybe something to the effect of

(define orig_ld_lib_path (getenv "LD_LIBRARY_PATH"))

(setenv "LD_LIBRARY_PATH"
(string-append "shared_obj" "/" (getenv
"GNUSTEP_HOST_CPU") "/" (getenv "GNUSTEP_HOST_OS") "/"
(getenv "LIBRARY_COMBO") ":" (getenv
"LD_LIBRARY_PATH")))

(dynamic-link "myClasses.so")
(setenv "LD_LIBRARY_PATH" orig_ld_lib_path)

>I also had to add (use-modules (languages
gstep->guile)) at the beginning
>of each script, shouldn't Greg add this by default to
>its scripts
>.without having to do this by hand ?

I noticed this too, yesterday, when trying to run an
individual test where (use-modules (languages
gstep-guile)) was called in begin.grg 

only thing i can think of against having greg add this
by default is maybe it's supposed to be generic so it
could be usable by non-gnustep stuff?

I was thinking we could add something like (if
(file-exists? "common.grg") (load "common.grg")) to
greg somewhere if not the (use-modules (...

and put stuff like use-modules and maybe other stuff
common to all tests ./ in common.grg so we wouldn't
have to duplicate it in every script.. 

matt

__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com