MapServer going mobile on Zaurus

[blaise]

Le dimanche 27 Mars 2005 11:15, blaise a écrit :
> Le vendredi 25 Mars 2005 07:22, blaise a écrit :
> > Le vendredi 25 Mars 2005 05:32, blaise a écrit :
> > > Hi,
> > > I would have like to post this on the devel list but I can't. (I'm
> > > often lost with sucribtion proccess...sorry)
> > >
> > > Well, I hope I'm close to it. I got a Zaurus sl-5500 recently to try to
> > > put MapServer inside.
> > >.........
> > >.........
> > >  MapServer finds the libs at configure and build time. I am still
> > > missing the libtiff devel package, but I should get it or I will try to
> > > build it. The configure pass seems to go smoothly, I can join the
> > > output if needed; but I get a strange kind of errors at build :
> > >
> > > ./libmap.a(mapstring.o): In function 'trimLeft':
> > > mapstring.o(.text+0x258): undefined reference to '__ctype_b'
> > > ./libmap.a(mapstring.o): In function 'MSEncodeUrl':
> > > mapstring.o(.text+0x98c): undefined reference to '__ctype_b'
> > > ./libmap.a(mapstring.o): In function 'hash':
> > > maphash.o(.text+0x98): undefined reference to '__ctype_tolower'
> > > collect2: ld returned 1 exit status
> > > make: *** [shp2img] Error 1
> >
> > Now it build fine !
> > I found a trick in a kind of RedHat bugzilla. But fully applied, it gave
> > other errors, so I just kept what was needed... I'ld like to know it what
> > I did could corrupt MapServer :
> > In both files mapstrinc.c and maphash.c, I added a line at the very
> > begining, just after this one :
> > #include <ctype.h>
> >
> > For mapstring.c, I added this line :
> > __const unsigned short int *__ctype_b;
> > and for maphash.c, this one :
> > __const __int32_t *__ctype_tolower;
> >
> > Can this mess up MapServer ?
>
> Not yet working :-( and quiet fastidious to setup the whole thing...
> Puting Mapserver cgis in the apache cgi-bin directory and pointing at :
> http://localhost/cgi-bin/mapserv?
> gives :
> "No query information to decode. QUERY_STRING is set, but empty."
>
> MapServer infos gives :
> #mapserv -v
> MapServer version 4.4.1 OUTPUT=PNG OUTPUT=JPEG OUTPUT=WBMP
> SUPPORTS=FREETYPE INPUT=EEPL7 INPUT=JPEG INPUT=SHAPEFILE
>
> But initializing Itasca demo gives :
> "loadMapInternal(): General error message. Given map extent is invalid."
> Quite uncommon mapserver error, isn't it ?
>
> Apache error log file tells :
> [Desktop Entry]
> [...Date and time...] [warn] pid file /var/run/apache.pid overwritten --
> Unclean shutdown of previous Apache run?]
> [...Date and time...] [notice] Accept mutex: sysvsem (Default: sysvsem)]
> [...Date and time...05:23] [notice] Apache/1.3.29 (Debian GNU/Linux)
> configured -- resuming normal operation]
>
> I divided by ten the main and reference images size in demo.html and
> demo.map. Also removed the decimal of the main and reference images EXTENT.
> Does the reference image have to be of one of MapServer image INPUT types ?
>
> How can I debug ? I have never used MapServer log files, could it be of any
> use ?
> Do you think the RedHat trick I used to allow MapServer to build (previous
> post) is responsible for that error ? Or is it really some differences
> between gcc-2.95.2 and gcc-3.xx ?
> This error is hard coded in mapfile.c and mapobject.c files. I am going to
> check these files, see also if I find a better trick that could allow a
> nice MapServer building, and try to slim Itasca demo step by step to see
> what is working and what is not.
> One last thing : are bison and flex needed to compile nicely the basic
> MapServer cgi ? I have bison but it is not clear if flex is well installed
> or not.
> Any ideas for debuging welcomed. Thks.
> Blaise

I tested with MapServer-3.6.0 and itasca-3.5 demo.
Same error on build, fixed with the (bad?) RedHat trick, then build fine.
http://localhost/cgi-bin/mapserv? and #mapserv -v are ok, but Initializing the 
demo gives :
"Internal Server Error"
Apache error log file doesn't tell anything.
Where could I found an older MapServer version ? Anyway, I don't think the 
problem lives here, it must be this c-code RedHat tricks, or a misconfig of 
apache. . . ?