Peter Prymmer writes:
> Charles Lane wrote:
>> Here's a new version of the CONFIGURE.COM patch, fixing the typo
>> Craig found and revising the myread routine....with a bunch of ancillary
>> changes to take advantages of the revised myread.
> >
>> Plus a couple of fixes: stuff in config.sh that was set to 'Y'/'N'
>> instead of 'define'/'undef'; and removing 000000 from install prefix;
>> removing prompt for "OSNAME" when we won't accept any changes anyway.
> OK there go all the little jokes ... sigh.
I'll try and put some more jokes in...maybe ask about the airspeed of
an unladen swallow if you select a particularly boneheaded config.
>> +$ echo4 "(OK, will run with -d after this question.)"
> The line you replaced read:
>> -$ echo4 "(OK, I will run with -d after this question.)"
> ^^^
> so perhaps there ought to be an "I" in there. Those prompts
> were taken directly from the sh version of Configure by the
> way.
Ah, okay...anthropomorphism restored.
>> -$ DELETE/SYMBOL config_dflt
>> +$ if f$type(config_dflt) .nes. "" DELETE/SYMBOL config_dflt
> ^^^
> This seems to be missing a THEN, as in:
Got it, good catch!
>> -Configure uses the operating system name and version to set some defaults.
>> -The default value is probably right if the name rings a bell. Otherwise,
>> -since spelling matters for me, either accept the default or answer "none"
>> -to leave it blank.
> Actually perhaps myread ought to be able to handle the string "none".
> I think that the sh version does.
Added to myread processing.
>> $ prefixbase = prefix - "]"
>> $! Add _ROOT to make install PERL_ROOT differ from build directory.
>> $ prefix = prefixbase + "_ROOT.]"
> There is another problem right in this vicinity (of my doing recently).
> I thought that I had tested the F$GETDVI() with some non sense input
> but note this result:
> $ sho dev disk$nonesuch
> %SYSTEM-W-NOSUCHDEV, no such device available
> $ @configure "-Dusedevel" "-Dprefix=disk$nonesuch:[perl5_7_2.]" "-des"
> First let's make sure your kit is complete. Checking...
> Looks good...
> Shall I @ [-]config.sh for default answers? [n]
> Operating system name? [VMS]
> (-18000) Is this UTC Time Zone Offset correct? [y]
> What is your architecture name? [VMS_AXP]
> Will you be sharing your PERL_ROOT with a VAX? [n]
> Installation prefix to use (for PERL_ROOT)? [ disk$nonesuch:[perl5_7_2.] ]
> %SYSTEM-W-NOSUCHDEV, no such device available
> \disk$nonesuch:[perl5_7_2.]\
> %DCL-E-INVIFNEST, invalid IF-THEN-ELSE nesting structure or data inconsistency
> $
> I think that configure.com ought to allow for the setting up of a build
> on volumes that may not yet be mounted, since it allows the perl admin
> to mount after running `mms test` or whatever they want. The call to
> F$GETDVI(perl_root,"MNT") was not intended to be a fatal error, but
> merely a reminder that the device was not mounted. Perhaps I had not
> tested it on VMS 7.3, perhaps I neglected to test properly. At any
> rate this sort of change should perhaps be incorporated into your patch
> (since you've been tweaking configure.com quite a bit lately it might
> put less wear and tear on the pumpking to see one unified patch):
Patch applied.
> I hope that could be incorporated. Even better would be setting an appropriate ON
> handler and uncommenting the attempt to CREATE/DIRECTORY and warning of inability
> (a simple writability test as it were) :-)
Okay, I'll take a look at implementing this. I presume one would want
to have a prompt ("do you want to create this directory...") before
the creation attempt.
I've also been cooking up some DCL to read in the CONFIG.SH and set
appropriate DCL symbols, since just doing @CONFIG.SH isn't really that
useful on VMS. The code to read CONFIG.SH is pretty simple; it's
making sure that stuff we want as future defaults is actually saved to
CONFIG.SH that is tricky.
--
Drexel University \V --Chuck Lane
======]---------->--------*------------<-------[===========
(215) 895-1545 _/ \ Particle Physics
FAX: (215) 895-5934 /\ /~~~~~~~~~~~ [EMAIL PROTECTED]
