Hi,
Just a short note to say that I've not had much luck compiling our software with the two 2003 gpc alpha releases. Everything works fine with the 2.1 release.
Running gpc20030209 built with the gcc 3.2.1 backend, I got a segmentation fault whilst compiling one of our modules (no prior errors/warnings to indicate what may have caused it).
Running gpc20030323, again built with the gcc 3.2.1 backend, the segmentation fault goes away but the compiler now hangs in an endless loop when I compile any module that uses a particular unit I have. Doing --debug-source seems to show the compiler hanging part way through each module, but I haven't found a pattern to the code which causes it to stop.
I've now run the gpc20030323 i686 binary (built with gcc 2.95.3) and I'm getting a new gpc1 segmentation fault. This occurs when I do
CASE i OF 0: Segmentation fault here BEGIN ... END;
It occurred because integer i was no longer defined. The declaration was the no longer valid i : static integer which needed to be i : integer attribute (static); Admittedly, fixing this cured the fault and it appears I may be able to compile everything with this binary (I still have a lot of changes to make). However, a segmentation fault is a nasty way of exiting the compiler.
I will do the code conversions to compile with this release (which you're right - I do hate) but the gcc 3.2.1 backend builds still seem fragile.
Sorry I can't be any more specific about the bugs yet and I don't want to sound critical. I think gpc is in great shape thanks to all your efforts. The gcc3 step is no easy one to take as we have found in many other areas.
Regards
David Wood QinetiQ Farnborough
The Information contained in this E-Mail and any subsequent correspondence is private and is intended solely for the intended recipient(s). For those other than the recipient any disclosure, copying, distribution, or any action taken or omitted to be taken in reliance on such information is prohibited and may be unlawful.
Wood David wrote:
Running gpc20030323, again built with the gcc 3.2.1 backend, the segmentation fault goes away but the compiler now hangs in an endless loop when I compile any module that uses a particular unit I have. Doing --debug-source seems to show the compiler hanging part way through each module, but I haven't found a pattern to the code which causes it to stop.
You might be able to see more in the debugger. (Either run the compiler within gdb, or in this case it might be easier to attach gdb to the running compiler to see where it's looping.)
I've now run the gpc20030323 i686 binary (built with gcc 2.95.3) and I'm getting a new gpc1 segmentation fault. This occurs when I do
CASE i OF 0: Segmentation fault here BEGIN ... END;
It occurred because integer i was no longer defined. The declaration was the no longer valid i : static integer which needed to be i : integer attribute (static); Admittedly, fixing this cured the fault and it appears I may be able to compile everything with this binary (I still have a lot of changes to make). However, a segmentation fault is a nasty way of exiting the compiler.
Sure. Easy fix. Still, I'd *really* appreciate to get a complete test program, even if it's only a few lines!
Frank
-
Frank Heckenbach -
Wood David