June 2002 - Cvs-hugs - Haskell.org

cvs commit: hugs98/tests/ffi .cvsignore
by Alastair Reid 20 Jun '02

20 Jun '02

reid 2002/06/20 09:10:31 PDT Modified files: tests/ffi .cvsignore Log: More thoroughly ignore intermediate files Revision Changes Path 1.2 +3 -1 hugs98/tests/ffi/.cvsignore

1 0

RE: cvs commit: hugs98/src HsFFI.h builtin.c connect.h ffi.c static.c
by Simon Marlow 19 Jun '02

19 Jun '02

> reid 2002/06/17 14:46:48 PDT > > Modified files: > src HsFFI.h builtin.c connect.h ffi.c > static.c > Log: > Made dynamic and wrapper forms work. > > This is my current test program: Thanks Alastair, I slurped your test program into GHC's test suite. I had to make a few changes to get it to run: the main one is that you used 'Ptr Char' in a lot of places where the type should have been 'Ptr CChar', i.e. CString. I guess in Hugs that doesn't matter too much because Char and CChar have the same representation at the C level, but that's not the case for GHC. I've attached my version. You might want to take a look in fptools/testsuite/tests/ghc-regress/ccall, and see if there's anything in there that might be useful for your test suite. Cheers, Simon

1 0

cvs commit: hugs98/src/unix aclocal.m4 configure configure.in
by Alastair Reid 18 Jun '02

18 Jun '02

reid 2002/06/18 15:40:19 PDT Modified files: src/unix aclocal.m4 configure configure.in Log: Stole some autoconfery from fptools/{aclocal.m4,configure.in}. Thanks Dudes! Revision Changes Path 1.6 +191 -0 hugs98/src/unix/aclocal.m4 1.23 +4205 -159 hugs98/src/unix/configure 1.26 +85 -0 hugs98/src/unix/configure.in

1 0

cvs commit: hugs98/lib/exts Storable.hs
by Alastair Reid 18 Jun '02

18 Jun '02

reid 2002/06/18 13:28:25 PDT Modified files: lib/exts Storable.hs Log: Replaced libspec "[Storable_aux.o]" with "[Storable_aux.c]" Revision Changes Path 1.4 +51 -51 hugs98/lib/exts/Storable.hs

1 0

cvs commit: hugs98/tests testScript hugs98/tests/ffi FileIO.hs FileIO.input FileIO.output String.hs String.input String.output
by Alastair Reid 18 Jun '02

18 Jun '02

reid 2002/06/18 13:26:55 PDT Modified files: tests testScript Added files: tests/ffi FileIO.hs FileIO.input FileIO.output String.hs String.input String.output Log: More runtime checks. I'm not sure how portable these are - I expect we'll iterate a few times to find something portable. You should expect to see some spurious warnings from gcc as it complains about perfectly legal but unusual code (like giving a function prototype for gcc builtin functions). The output I see from these tests is: ./../src/hugs +q +w -pHugs: -P../lib:../lib/hugs:../lib/exts +G ffi/Sin.hs < /dev/null expected stderr not matched by reality *** /tmp/no_stderr27731 Tue Jun 18 21:34:24 2002 --- /tmp/runtest27731.2 Tue Jun 18 21:34:24 2002 *************** *** 0 **** --- 1,2 ---- + Sin.c: In function `hugsprim_sin_3': + Sin.c:51: warning: built-in function `sin' declared as non-function -- !!! Testing static, dynamic and wrapped import of trig function -- !!! Testing marshalling of strings ./../src/hugs +q +w -pHugs: -P../lib:../lib/hugs:../lib/exts +G ffi/String.hs < /dev/null expected stderr not matched by reality *** /tmp/no_stderr27767 Tue Jun 18 21:34:25 2002 --- /tmp/runtest27767.2 Tue Jun 18 21:34:25 2002 *************** *** 0 **** --- 1,3 ---- + MarshalUtils.c: In function `hugsprim_memcpy_0': + MarshalUtils.c:22: warning: conflicting types for built-in function `memcpy' + MarshalUtils.c:22: warning: extern declaration of `memcpy' doesn't match global one -- !!! Testing marshalling of strings -- !!! Testing system calls ./../src/hugs +q +w -pHugs: -P../lib:../lib/hugs:../lib/exts +G ffi/FileIO.hs < /dev/null expected stderr not matched by reality *** /tmp/no_stderr27827 Tue Jun 18 21:34:25 2002 --- /tmp/runtest27827.2 Tue Jun 18 21:34:26 2002 *************** *** 0 **** --- 1,3 ---- + MarshalUtils.c: In function `hugsprim_memcpy_0': + MarshalUtils.c:22: warning: conflicting types for built-in function `memcpy' + MarshalUtils.c:22: warning: extern declaration of `memcpy' doesn't match global one -- !!! Testing system calls Revision Changes Path 1.3 +8 -0 hugs98/tests/testScript

1 0

cvs commit: hugs98/tests .cvsignore testScript hugs98/tests/ffi .cvsignore Sin.hs Sin.input Sin.output
by Alastair Reid 18 Jun '02

18 Jun '02

reid 2002/06/18 13:02:08 PDT Modified files: tests .cvsignore testScript Added files: tests/ffi .cvsignore Sin.hs Sin.input Sin.output Log: Simple runtime checks Revision Changes Path 1.2 +1 -1 hugs98/tests/.cvsignore 1.2 +4 -0 hugs98/tests/testScript

1 0

cvs commit: hugs98/tests/ffi types_aux.c
by Alastair Reid 18 Jun '02

18 Jun '02

reid 2002/06/18 12:44:28 PDT Added files: tests/ffi types_aux.c Log: forgeto

1 0

cvs commit: hugs98/src/unix configure configure.in hugs98/tests config.in testScript Readme testScript.in
by Alastair Reid 18 Jun '02

18 Jun '02

reid 2002/06/18 12:37:54 PDT Modified files: src/unix configure configure.in tests Readme Added files: tests config.in testScript Removed files: tests testScript.in Log: Important change to testscript - read this if you are adding/modifying tests. I have tweaked testscript so that it consists of: 1) A small variable part config.in 2) A large fixed part testScript I have removed testScript.in and tweaked configure to modify the small variable part. You no longer have to remember to copy changes you're testing in testScript over into testScript.in. Revision Changes Path 1.22 +16 -12 hugs98/src/unix/configure 1.25 +2 -2 hugs98/src/unix/configure.in 1.6 +3 -9 hugs98/tests/Readme

1 0

cvs commit: hugs98/tests testScript.in hugs98/tests/ffi Forms.hs Loaded.output Types.hs forms_aux.c forms_aux.h
by Alastair Reid 18 Jun '02

18 Jun '02

reid 2002/06/18 12:14:39 PDT Modified files: tests testScript.in Added files: tests/ffi Forms.hs Loaded.output Types.hs forms_aux.c forms_aux.h Log: First cut at a test suite for the ffi. As you'd expect you do 'sh ./testScript ffi' to run the test. These checks only check that typechecking and code generation works ok for correct input. So far, there are no checks for erroneous input and there are no checks that it executes ok. Revision Changes Path 1.23 +15 -3 hugs98/tests/testScript.in

2 1

RE: cvs commit: hugs98/src ffi.c
by Simon Marlow 18 Jun '02

18 Jun '02

> > btw The storable instances are bogus - fixing now > > Ooops, they were in fact fine (I got confused over whether they should > use byte offsets or element offsets). > > But while killing my commit of the 'fixed' code I got myself > in this state: > > cvs server: [09:47:46] waiting for reid's lock in > /home/cvs/root/hugs98/lib/extscvs server: [09:48:16] waiting > for reid's lock in /home/cvs/root/hugs98/lib/extscvs server: > [09:48:46] waiting for reid's lock in > /home/cvs/root/hugs98/lib/extsKilled by signal 2. > > I have no idea how to clean this up. I assume you've probably fixed this by now. For future reference (and anyone else who runs into this problem): stale locks tend to be left around when CVS crashes, and have to be removed by hand by logging on to cvs.haskelll.org with SSH, and removing the appropriate lock file from /home/cvs/root/<wherever> which is usually called something beginning with '.#cvs' IIRC. Cheers, Simon

2 1