home *** CD-ROM | disk | FTP | other *** search
-
-
- This makefile appears to work after enough tries that I thought I was
- going to do the 100,000 monkeys thing all by myself.
-
- Assuming a directory tree of
-
- /GS5.10
- /jpeg-6a /libpng-0.96 /zlib-1.0.4
-
- with all GS source in GS5.10 and an assign to it, here is the current
- build_gs script:
- ----------------
- assign ghostscript: ghost:ghostscript/gs5.10
- assign GSBIN: Ghost:ghostscript/5.10
- date >GSBIN:Build_All_Start
- delete gs_000 gs_000.1 gs_020 gs_020.1 gs_040sf gs_040fpu
- IF NOT EXISTS gs_000
- echo Building gs_000
- make clean
- date >GSBIN:Build_000_begin
- make "FPU_TYPE=0" "GS=gs_000" "XLDFLAGS=-noixemul" "XCFLAGS=-m68000"
- strip gs_000
- copy gs_000 GSBIN:
- date >GSBIN:Build_000_Done
- ENDIF
- IF NOT EXISTS gs_020.1
- echo Building gs_020
- make clean
- date >GSBIN:Build_020_Start
- make "FPU_TYPE=1" "GS=gs_020" "XLDFLAGS=-noixemul" "XCFLAGS=-m68020"
- strip gs_020
- copy gs_020 GSBIN:
- date >GSBIN:Build_020_Done
- ENDIF
- IF NOT EXISTS gs_020fpu
- echo Building gs_020fpu
- make clean
- date >GSBIN:Build_020fpu_Start
- make "FPU_TYPE=2" "GS=gs_020fpu" "XLDFLAGS=-noixemul" "XCFLAGS=-m68020 -m68881"
- strip gs_020fpu
- copy gs_020fpu GSBIN:
- date >GSBIN:Build_020fpu_Done
- ENDIF
-
- ;make "FPU_TYPE=1" "GS=gs_030" "XLDFLAGS=-noixemul" "XCFLAGS=-m68030"
- ;strip gs_030
- ;copy gs_030 Ghostscript:
- ;make clean
-
- ;make "FPU_TYPE=2" "GS=gs_030fpu" "XLDFLAGS=-noixemul" "XCFLAGS=-m68030 -m68881"
- ;strip gs_040fpu
- ;copy gs_030fpu Ghostscript:
- ;make clean
-
- ; Note, the gs_040sf code generated is identical to the gs_020 code, so
- ; there isn't much use doing it twice!
- ;
- ;IF NOT EXISTS gs_040sf
- ; echo Building gs_040sf
- ; make clean
- ; date >GSBIN:Build_040sf_Start
- ; make "FPU_TYPE=1" "GS=gs_040sf" "XLDFLAGS=-noixemul" "XCFLAGS=-m68040 -msoft-float"
- ; strip gs_040sf
- ; copy gs_040sf GSBIN:
- ; date >GSBIN:Build_040sf_Done
- ;ENDIF
- IF NOT EXISTS gs_040fpu
- echo Building gs_040fpu
- make clean
- date >GSBIN:Build_040fpu_Start
- make "FPU_TYPE=2" "GS=gs_040fpu" "XLDFLAGS=-noixemul" "XCFLAGS=-m68040 -m68881"
- strip gs_040fpu
- copy gs_040fpu GSBIN:
- date >GSBIN:Build_040fpu_Done
- ENDIF
- make clean
- date >GSBIN:Build_All_Done
- ---------------------
-
- What I found that fixed the pathnames problem was partially my fault! When
- I couldn't get it to work, and you have that copy if its not been deleted,
- I started through the makefile adding comments as to what the 'strcat'ing
- should have made out of the path at that point in the file.
-
- Which was neat I thought. Silly me.
-
- Once I started comparing what I was getting in the object to what I had in
- the makefile, two things jumped out at me. The first being that the :
- character is apparently the make/gcc end of pathlist marker, and it
- automaticly inserts a space after the : before continueing on with the rest
- of the pathlist.
-
- Now, I can't find it in any of the docs, but I litteraly dreamed something
- about using single quotes around such strings in order to hide them from
- that rule, wherever it is. I had to do some experimenting to see what
- actually worked, and the string that needs protected is apparently those in
- the $(reference) style, requiring that they be '$(reference)' when used.
- This is for
-
- reference = Ghostscript:
-
- Ok, that was high jump #1. I have used a binary editor, FileX2.0, to
- adjust these strings in the object until they did work, but thats a hell of
- a way to run a railroad IMO.
-
- So after getting a good compile for gs_000 except for the fact that a good
- looking set of search paths in the --help screen didn't = working, It can
- find its fonts code.
-
- So the binary editor gets called out for overtime again. And I fixed it
- again so it worked, this time by zeroing out a string of tabs (0x09's)
- after the default pathlists, which were displayed spread out all over the
- width of a picasso screen.
-
- It finally glommed on me that the tabs I was using to tab over to the
- comment character column were being included as part of the pathlist strings!
-
- Sooo, my comments hit the bin, sort of. They got moved to the line above
- and a return is now the next character after the last character of the
- pathlists. The script above has just now been restarted, so it'll be a few
- hours before I know that this positively worked.
-