Jon Mountjoy writes:
>
...
>
> Perl had problems with the ghc script. I tried vs. 5 and 4 to no
> avail. In the ghc script there are the lines
>
> if $StkChkByPageFaultOK;
> && $TargetPlatform !~ /nextstep/;
> .....
> if $StkChkByPageFaultOK;
> && $TargetPlatform !~ /nextstep/;
> # I do not know how to do STACK_CHECK_BY_PAGE_FAULT
> # on NeXTs (my fault, not theirs), so I don't.
> # CaS
> &&
>
> The semicolons seem to be causing the hiccup. If I remove them (all
> those before an &&), then it seems to work fine. Is this a bug and is
> this a fix?
>
This is a bug and a fix. I naughtily yanked the original source
distribution to fix this embarrassing problem just after the 2.02
release went out - did you pick up the source distribution just now?
> ######### 2
> ...
> ------------------------------------------------------------------------
> ../../../literate/mkdependlit -o .depend user.lit
> Giant error 'do'ing lit-globals.prl: at ../../../literate/mkdependlit line 636.
> gmake[3]: *** [depend] Error 2
> gmake[2]: *** [boot] Error 2
>
> Is this also due to a broken Perl?
>
Sigh, I'm leaning towards a yes here (it appears to be dying when doing a
random require()) - does this happen with both perl4 & perl5? Here's a
wild shot though - try deleting the comment line that ends
lit-globals.prl..
> ########### 3
>
> Somewhere else (it might be in the gmake all now), you call
> mkdependHS-1.2 Now I *think* that this only gets installed as
> mkdependHS with the result that the binary cannot be found.
>
> ########## 4
>
Have a look in the install guide/release notes on the new Makefile
setup, you currently cannot boot the compiler using 2.02 (it's on our
ToDo list...), but need the Good Old Compiler (0.29). Hence, the
mkdependHS script the Makefile is pointing at. There's a make variable,
MKDEPENDHS_1_2, controlling which script to invoke - see mk/config.mk.in
Hope that helps,
--Sigbjorn
