After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 69809 - compilation fails using sun's workshop cc
compilation fails using sun's workshop cc
Status: VERIFIED FIXED
Product: libxslt
Classification: Platform
Component: general
1.0.10
Other Solaris
: Normal blocker
: ---
Assigned To: Daniel Veillard
Daniel Veillard
Depends on:
Blocks:
 
 
Reported: 2002-01-27 12:47 UTC by Michael Kroell
Modified: 2009-08-15 18:40 UTC
See Also:
GNOME target: ---
GNOME version: ---



Description Michael Kroell 2002-01-27 12:47:14 UTC
compiling with "CC=cc CFLAGS="-xO3 -xdepend" ./configure --without-debug" 
croaks with '"transform.c", line 3282: syntax error before or at: }'

the following patch against the current cvs-cersion helps as a workaround:
<snip>
--- sib/test/libxslt/libxslt/transform.c        Tue Jan 15 10:41:46 2002
+++ sie/test/libxslt-1.0.10/libxslt/transform.c Sun Jan 27 13:27:01 2002
@@ -3279,6 +3279,9 @@
     xsltGenericDebug(xsltGenericDebugContext,
             "xsltApplyStripSpaces: removed %d ignorable blank node\n", 
nb);
 #endif
+#ifdef __SUNPRO_C
+1;
+#endif
 }
 
 /**
</snip>
I am no C-Programmer (as you see looking at that patch;-). Does the 
Workshop Compiler need a statement after the label-declaration or before 
the end of the function? don't know...but it compiles with that and is 
*significantly* faster at the test-results as the gcc-compiled version :/
Comment 1 Daniel Veillard 2002-01-27 12:53:33 UTC
heh :-)

Okay the clean way to do this is:

-------------------
done:
#ifdef WITH_XSLT_DEBUG_PROCESS
    xsltGenericDebug(xsltGenericDebugContext,
             "xsltApplyStripSpaces: removed %d ignorable blank
node\n", nb);
#endif
    return;
}
--------------------

  The return should not have any side effect, and is perfectly
portable, thanks for the information.

BTW what kind of improvements are you seeing ? Maybe you should
get in touch with Garry Pennington who mainatins the Solaris
binaries (he's also  a Sun Microsystems employee).

Daniel
Comment 2 Michael Kroell 2002-01-28 09:14:15 UTC
Daniel, merci beaucoup for the amazingly quick response!

>BTW what kind of improvements are you seeing ? Maybe you should
>get in touch with Garry Pennington who mainatins the Solaris
>binaries (he's also  a Sun Microsystems employee).
I'll point him there :-)

Following, test results of the the two compilations

Enterprise 250 2x400 (gcc -O2)
c102mk@sie:/mp_sie2/ci00/build/sie/libxslt-
1.0.10gcc_slow/tests/xmlspec $ cat debug
Parsing stylesheet ./REC-xml-2e.xsl took 3 ms
Parsing document ./REC-xml-20001006.xml took 312 ms
Applying stylesheet took 1191 ms
Saving result took 82 ms

XML::LibXSLT-Test:
Files=8,  Tests=69, 15 wallclock secs ( 1.99 cusr +  0.49 csys =  
2.48 CPU)

Enterprise 250 2x400 (workshop5 -xO3 -xdepend)
c102mk@sie:/mp_sie2/ci00/build/sie/libxslt-1.0.10/tests/xmlspec $ cat 
debug
Parsing stylesheet ./REC-xml-2e.xsl took 3 ms
Parsing document ./REC-xml-20001006.xml took 252 ms
Applying stylesheet took 1150 ms
Saving result took 82 ms

XML::LibXSLT-Test:
Files=8,  Tests=69,  3 wallclock secs ( 2.08 cusr +  0.39 csys =  
2.47 CPU)

With "significantly" I refered to the performance differences (15wcs 
to 3wcs) of the XML::LibXSLT tests. You could not oversee the that 
the gcc compile was unusual slow in AxKit action :/
On the other hand, this is the only box I tested it, so I don't have 
comparable references.

Michael

P.S.:
Out of curiousity I looked at the test-results at an Intel-Box
Pentium III (Coppermine) 1000 (gcc)
[c102mk@cie xmlspec]$ cat debug 
Parsing stylesheet ./REC-xml-2e.xsl took 4 ms
Parsing document ./REC-xml-20001006.xml took 82 ms
Applying stylesheet took 396 ms
Saving result took 34 ms
Comment 3 Daniel Veillard 2002-01-28 10:40:55 UTC
Okay thanks, seems the difference doesn't come from
libxslt performances itself, strange.

  thanks again,

Daniel
Comment 4 Daniel Veillard 2002-02-12 08:46:20 UTC
Okay this should be fixed in libxslt-1.0.12,

 thanks !

Daniel