Achim, Thank you for sending those patches. I've modified them a bit and applied them in CVS. I had a go at building the portable version myself with the latest version (4.1.4) of the driver and actually found some problems which I've now fixed in CVS. I hadn't considered building an interpreted database for the latest version since I wasn't aware that anyone was using it. portDB.txt on www.polyml.org was built with version 4.0 and I'd assumed that the interpreted version was only going to be used as a porting aid. The code-generator for the portable version will need to be modified to allow it to build version 4.1.4. I'll have a look at this but until I do I won''t know how much work is involved. Are there many people using the interpreted version?
David.
Achim D. Brucker wrote:
Hi, congratulations to this new release candidate!
While testing this release, I recognised that the the interpreted (CPU independent) version of PolyML and and the readport program do not compile. Also, there is no portable database available for that release. Nevertheless I tried to fix the compile errors. Attached to this e-mail you will find a small patch which hopefully fixes this. This patch should apply cleanly to the PolyML 4.1.1 sources (from sf.net).
I tested my patch with the following configurations:
- GNU/Linux on a i386. The interpreted version can be build after disabling the i386 check in the configure script. Also readport builds without problems. I could successfully convert the old portDB (from www.polyml.org) with readport and run the interpreter.
- GNU/Linux on a XScale/StrongArm. Both, readport and the interpreter can be built without problems. Also converting the portDB from the old release and running the interpreter works.
I also started to eliminate all the warning that occur during compiling the driver (trying to fix the byte/char and unsigned/signed pointer assignments). If you like, I can probably send you a patch for this in the next days.
Achim