Re: PSPP & PSPPIRE running as native windows applications.

[John McCabe-Dansted]

On 2/7/07, John Darrington <address@hidden> wrote:
> Three of the tests were failing merely due to whitespace differences.
> I've fixed these three tests.

I've done a cvs up and recompiled. Now I get 33 failures instead of just 32 :(

The new log is at:
http://www.csse.uwa.edu.au/~john/newmake2.log

> The differences shown in the  RANK test is interesting.  I suspect the
> problem is at src/language/stats/rank.q:701
>
>   snprintf(func_abb, 4, "%s", function_name[f]);
>
> according to my man page " The functions snprintf() and vsnprintf() do
> not write more than size bytes (including the trailing '\0').", but
> your version does not seem to include the trailing null in the size,
> so we're seeing one more character.  Perhaps you should report that
> to the maintainer of your C library.  Alternatively perhaps we should
> be using gnulib/autoconf to deal with broken implementations.

I made a testfile. Your suspicions are correct.
 i=snprintf(x,4,"12345");
 Linux: i=5, x="123"
 Wine: i=-1, x="1234"

This is a known bug:
 http://sourceforge.net/mailarchive/message.php?msg_id=38119968

Presumably wine follows the old incorrect windows behavior. I suspect
that we simply shouldn't rely on snprintf being correct on windows,
since testing for correctness of snprintf when cross-compiling doesn't
really tells us anything useful. Perhaps we should just bundle a
portable version of snprintf in the windows.exe, such as the
following:
 http://www.ijs.si/software/snprintf/

The bad snprintf might also cause segfaults if it fails to terminate
the string with '\0'.

--

On the topic of *printf, I had to comment out
 libintl_fprintf libintl_printf and libintl_sprintf
In intl/printf.c because these symbols conflict with symbols provided by -lintl

--

I suspect that "is_special_file_name" in src/data/file-name.c should
be replaced with "fn_is_special" as "is_special_file_name" does not
occur anywhere else in pspp or indeed the web.

--

I had to make the following changes to the Makefile so that the test
.exes would be built correctly:

<       tests/libpspp/ll-test \
<       tests/libpspp/llx-test \
<       tests/libpspp/heap-test \
<       tests/libpspp/abt-test
---
>       tests/libpspp/ll-test$(EXEEXT) \
>       tests/libpspp/llx-test$(EXEEXT) \
>       tests/libpspp/heap-test$(EXEEXT) \
>       tests/libpspp/abt-test$(EXEEXT)

--

configure added "-Dunix" to the Makefile. This did not seem good.
(Perhaps configure was picking up the host environment).

--

I defined some file permission flags that have no meaning under
windows to be 0, if they are undefined. Sys/wait is not implemented on
windows, so I didn't try to load the library. limits.h is needed on
windows to define INT_MIN and INT_MAX

Index: src/data/missing-values.h
===================================================================
RCS file: /sources/pspp/pspp/src/data/missing-values.h,v
retrieving revision 1.6
diff -r1.6 missing-values.h
79a80,87
> #ifndef S_IRGRP
> #define S_IRGRP 0
> #define S_IROTH 0
> #define S_IWGRP 0
> #define S_IWOTH 0
> #endif
80a89
>
Index: src/data/por-file-writer.c
===================================================================
RCS file: /sources/pspp/pspp/src/data/por-file-writer.c,v
retrieving revision 1.8
diff -r1.8 por-file-writer.c
106a107,114
> #ifndef S_IRGRP
> #define S_IRGRP 0
> #define S_IROTH 0
> #define S_IWGRP 0
> #define S_IWOTH 0
> #endif
===================================================================
RCS file: /sources/pspp/pspp/src/ui/debugger.c,v
retrieving revision 1.2
diff -r1.2 debugger.c
26c26
< #include <sys/wait.h>
---
> //#include <sys/wait.h>
38a39,40
>
Index: src/output/afm.c
===================================================================
RCS file: /sources/pspp/pspp/src/output/afm.c,v
retrieving revision 1.5
diff -r1.5 afm.c
35a36
> #include <limits.h>


--
John C. McCabe-Dansted
PhD Student
University of Western Australia