Re: Gasp - Some Simple Help Please

In article <3778AE22.C9120AD1@mindspring.com>, Tim Schaefer <tschaefe@mindspring.com> wrote: >> Tim Schaefer <tschaefe@mindspring.com> wrote: >> >This is exactly why Linux is not ready for prime time. I am also >> >disappointed in Informix for telling everyone that IDS 7.30 is >> >"certified" for 2.2.5 when you cannot even connect to the box. >> >> Who at Informix told you *that*? > >Hi Alan, head to this page please: > >http://www.informix.com/informix/products/linux/idslx.html Oh, yikes. I didn't know that page was changed. That's wrong -- a generic 2.2.5 (e.g. without the latest Alan Cox subrelease) will have problems with soctcp. I passed the word to get this fixed. >> The only 2.2 kernel implementation IDS 7.30 has been certified >> for is Red Hat 6.0, which needs NO patching whatsoever. If anyone >> told you otherwise, that person was mistaken. > >Hi Alan, head to this page please: >http://www.informix.com/informix/products/linux/idslx.html Like I said, the author of that paragraph was mistaken. >BTW >I managed to get things right by downloading and installing 2.2.10 from >the kernel.org. The following are working: > >* 4GL is working >* ESQLC is working With what compiler? egcs 2.91.XX ? gcc 2.7.X.X ? People have reported here that they get link errors with c4gl on Red Hat 6.0 (egcs 2.91.66) unless they install a backward-compatibility package and specify the older compiler (gcc something) via INFORMIXC. >> >IDS 7.3 - RPM forces you to use /opt/informix as the install directory. >> >> Wrong. That's just the *default*, following Linux standard (and as >> specified in the Filesystem Hierarchy Standard, I think). You can >> install it anywhere using: >> (For Red Hat 6.0+:) >> --relocate /opt/informix=/$INFORMIXDIR >> (For Red Hat 5.x:) >> --prefix=$INFORMIXDIR >> >> All of which is covered in the rpm man pages. > >Yeah, but Alan, I didn't see an example as I recall. I've also found >kpackage to be a really nice tool, to deal with RPMs, but if I'm not >mistaken you can't change the default--maybe you can with your options >above. I'm not familiar with kpackage, sorry. >The point is, as I stated before, RPM is completely unnecessary for Informix >installations. Notice how on AIX and HP there is NO RPM. How does one manage >without RPM on AIX, or HP? We use TAR! Hello! I'm not arguing with you; I'm saying that that's the kind of feedback you need to send to Informix directly. >I've since made my own TAR distributions of all the Linux software to avoid >having to mess with RPM. Now when I want to create a tree, I run a script that >untars software, goes through the install, and on to the next one. It's a >whole lot easier this way than even messing with RPM. Anyone that's any good >as an Informix professional can use TAR. It's actually easier to use than RPM, >and allows you all the control you need. I agree. RPM has it's strengths, but if you're not USING the other RPM benefits in your packaging scheme, it's not worth the disadvantages (especially the limitation of having one "copy" of a given package on the machine at a time without RPM griping). >why not take what I've said to the right >people, you being one of them, and take care of it. Maybe people who won't listen to me will listen to paying customers? -- Alan Denney yosemite at accesscom.com "Don't play with your food unless you're sure you can win." -- Thaves

Colonel Panic wrote: > >http://www.informix.com/informix/products/linux/idslx.html > > Oh, yikes. I didn't know that page was changed. That's wrong -- > a generic 2.2.5 (e.g. without the latest Alan Cox subrelease) > will have problems with soctcp. I passed the word to get this fixed. > Cool. Might as well mention 2.2.10 as the minimum. > >BTW > >I managed to get things right by downloading and installing 2.2.10 from > >the kernel.org. The following are working: > > > >* 4GL is working > >* ESQLC is working > > With what compiler? egcs 2.91.XX ? gcc 2.7.X.X ? > People have reported here that they get link errors with c4gl > on Red Hat 6.0 (egcs 2.91.66) unless they install a backward-compatibility > package and specify the older compiler (gcc something) via INFORMIXC. > My Environment: Linux Kernel 2.2.10 Distribution: SuSE 6.10 IDS 7.30.UC7 63048 Kybtes Single CPU Pentium III 500 Mhz w/192 MB RAM 18 GB SCSI Seagate HD Adaptec 2940 Ctlr $ gcc -v Reading specs from /usr/lib/gcc-lib/i486-linux/egcs-2.91.66/specs gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release) Doing a 4gl compile this is the tail end of a make: c4gl -O -globcurs -shared -nokeep -c calls08.4gl cc: -l4gsh720: linker input file unused since linking not done cc: /opt/informix/ids/lib/tools/libasf.a: linker input file unused since linking not done cc: -lcrypt: linker input file unused since linking not done cc: -lm: linker input file unused since linking not done --------------------------------- .4gl.o c4gl -O -globcurs -shared -nokeep -c calls07.4gl calls07.ec: In function `calls_rpt': calls07.ec:780: warning: comparison between pointer and integer cc: -l4gsh720: linker input file unused since linking not done cc: /opt/informix/ids/lib/tools/libasf.a: linker input file unused since linking not done cc: -lcrypt: linker input file unused since linking not done cc: -lm: linker input file unused since linking not done --------------------------------- .4gl.o c4gl -O -globcurs -shared -nokeep -c calls04.4gl cc: -l4gsh720: linker input file unused since linking not done cc: /opt/informix/ids/lib/tools/libasf.a: linker input file unused since linking not done cc: -lcrypt: linker input file unused since linking not done cc: -lm: linker input file unused since linking not done c4gl -O -globcurs -shared -nokeep calls01.o calls.o calls02.o calls05.o calls06.o calls08.o calls07.o calls04.o -o calls.4ge /bin/ls -l calls.4ge -rwxr-xr-x 1 informix informix 93551 Jun 30 06:32 calls.4ge strip calls.4ge /bin/ls -l calls.4ge -rwxr-xr-x 1 informix informix 78560 Jun 30 06:32 calls.4ge Testing my program, Queries work Inserts work Updates work Deletes work It's been a while since I messed with 4GL but all appears to be functioning correctly, except the spurious messages from the compile. 8-) ESQLC: 16 $ make -f calctime.mk ------------------------------------------- .ec.o esql -s -c calctime.ec 2>&1 calctime.ec: In function `main': calctime.ec:66: warning: passing arg 1 of `dttoasc' from incompatible pointer type esql -s calctime.o -o calctime /bin/ls -l calctime -rwxr-xr-x 1 informix informix 4428 Jun 30 06:41 calctime 17 $ ./calctime Usage: calctime "1997-04-01 06:03:28" "1997-04-17 09:03:57" where first value - is a starting time second value - is a stopping time 18 $ calctime "1997-04-01 06:03:28" "1997-04-17 09:03:57" 16 03:00:29 19 $ 19 $ c4gl -V INFORMIX-4GL Version 7.20.UD7 Software Serial Number AAC#A518469 20 $ esql -v Reading specs from /usr/lib/gcc-lib/i486-linux/egcs-2.91.66/specs gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release) /usr/lib/gcc-lib/i486-linux/egcs-2.91.66/collect2 -m elf_i386 -dynamic-linker /lib/ld-linux.so.2 /usr/lib/crt1.o /usr/lib/crti.o /usr/lib/g cc-lib/i486-linux/egcs-2.91.66/crtbegin.o -L/opt/informix/ids/lib -L/opt/informix/ids/lib/esql -L/usr/lib/gcc-lib/i486-linux/egcs-2.91.66 -L /usr/i486-linux/lib -lifsql -lifasf -lifgen -lifos -lifgls -lm -ldl -lcrypt /opt/informix/ids/lib/esql/checkapi.o -lifglx -lgcc -warn-common -lc -lgcc /usr/lib/gcc-lib/i486-linux/egcs-2.91.66/crtend.o /usr/lib/crtn.o /usr/lib/crt1.o(.text+0x36): undefined reference to `main' collect2: ld returned 1 exit status 21 $ esql -V INFORMIX-ESQL Version 9.16.UC3 Software Serial Number ACN#J266163 - > Alan Denney yosemite at accesscom.com > "Don't play with your food unless you're sure you can win." -- Thaves - -- --- Tim Schaefer ---- tschaefe@mindspring.com --- http://www.inxutil.com -- Informix on Linux Changed Everything - Check this out: http://www.it-director.com/99-06-29-4.html