jonnyg@ROVER.UMD.EDU (Jon Greenblatt) (03/04/88)
I started to compile the core.src on the IBM RT and found a problem. Add the the following lines to the begining of ./lib/X/Xlibint.h to fix the problem: #ifdef __STDC__ #undef __STDC__ #endif You may want to elaborate the #ifdef further if you are using more than one architecture for the source. The problem is that hc is an ansi like compiler but defines the __STDC__ define which is not ansi. This causes macro expansions in Xlibint.h to use ##'s instead of /**/'s as a macro spacer. Jon Greenblatt. (jonnyg@rover.umd.edu) (jonnyg@umd5.umd.edu)
alan@cunixc.columbia.edu (Alan Crosswell) (03/10/88)
I've run into a couple of other problems compiling on the RT. This is
on a system running the Pre-February release of 4.3. The pre-February
relase is called ACIS 4.3 while the February release (which you have to
pay for again) is called Academic Operating System 4.3. I wouldn't be
trying to compile X11 on the February release since it comes with it!
The hc man page says that although HC comes with an ANSI pre-processor,
the default is to use cpp. You have to give the +Hnocpp switch (or something
like that) to get the ANSI preprocessor.
Then, once you give up on getting HC to compile the toolkit (it complains
about not being able to resolve &NULL which is used in the XtOffset macro)
and you try to use pcc (by editing Rt.macros) you will find that X11/Xmd.h
has #if defined(ibm032) && !defined(_pcc_)
pragma blah blah blah
#endif
Well, cpp does not define _pcc_ anywhere. ibm032 does get defined. So,
you go back to Rt.macros again and add -D_pcc_ to STD_DEFINES (I guess).
Each of the three times I've tried to compile X10 and now that I'm compiling
X11, I've always tried to use HC since it is supposed to generate tighter
code. I've always gone back to ol' reliable slow horrible forgiving pcc
after fighting HC for a while. Maybe the February release of AOS 4.3 (sounds
like something that runs on a Cyber to me:-) has finally fixed HC enough to
really build X11.
If an when I get it built on my old, slow (still waiting on that CPU upgrade)
RT running ACIS 4.3 I will post a followup.
Alan Crosswell
ACIS 4.3 Site Coordinator
User Services
Columbia Universityjgm@K.GP.CS.CMU.EDU (John Myers) (03/10/88)
In article <8803031836.AA05252@rover.UMD.EDU> jonnyg@ROVER.UMD.EDU (Jon Greenblatt) writes: > I started to compile the core.src on the IBM RT and found a problem. >Add the the following lines to the begining of ./lib/X/Xlibint.h to fix the >problem: > >#ifdef __STDC__ >#undef __STDC__ >#endif May I suggest: #ifdef __HIGHC__ #undef __STDC__ #endif > You may want to elaborate the #ifdef further if you are using more > than one architecture for the source. The problem is that hc is an ansi > like compiler but defines the __STDC__ define which is not ansi. This > causes macro expansions in Xlibint.h to use ##'s instead of /**/'s as a > macro spacer. __STDC__ is a macro which is supposed to be defined if the compiler is an ANSI conforming compiler. hc is NOT an ANSI conforming compiler as it does not recognize the # and ## preprocessor commands. As the documentation for hc claims that it is an ANSI conforming compiler, I have APAR'ed this. -- John G. Myers jgm@k.gp.cs.cmu.edu
jonnyg@ROVER.UMD.EDU (Jon Greenblatt) (04/21/88)
I just compiled X11R2 on my RT PC. It was not easy but the changes were straight forward. From previous correspondence I learned that R2 on the RT has a lot of problems, does anyone know if the binary distributed on the Feb2 tape is more or less reliable? What was distributed was a stripped version of X11R1, I got the x10tox11 protocal converter to work. Switching between 6155 and ega displays is still buggy but the bugs are slightly different. As far as I can tell R2 works much better than the R1 on the binaries that came on the Feb2 release from IBM. If you would like a detailed explanation of how I got R2 up, send me mail, I will let the requests pile up so please be patient. The fixes I sent out a long time ago were incorrect, I did not have the correct update to 4.3 then. Remember, I did get R2 up but the xperts say it's not perfect! Well I think I'm going to run R2 for now on, so much for IBM binaries! Comments are welcome..... Jon Greenblatt. (jonnyg@rover.umd.edu)