[comp.lang.perl] Casts in h2ph PLUS Sequent Problems

rbj@uunet.UU.NET (Root Boy Jim) (02/13/91)

In article <121466@uunet.UU.NET> rbj@uunet.UU.NET (Root Boy Jim) writes:
>Inside sizeof is the only place I can see any use for casts.
>Actually, I'm pretty amazed at your translations.
>
>MICHAEL FINGERHUT, THIS MAY BE YOUR PROBLEM!

It may be, but then maybe not. I have serious problems with
TIOCSETP on a Sequent S81 Symmetry running Dynix 3.0.17.9.
I sometimes get "improper operation for device" (or similar),
or I get a floating point exception on TC's tcbreak script
in the h2pl directory. Both of these work on Pyramid and Sun.
I get the same results whether or not I use Sequent's original
or suped up compiler (ATSCC), with or without -O or -g.

Larry, would you like a temporary account to fix this?

Also, Sequent has a parallel make facility. If you have a line like

	target: & dependents

then dependents will be made in parallel. Quite handy when you
have twenty processors at your disposal. Sequent's make seems
to have macro '&' defined to '&' so using '$&' should cover
any system. I think Encore has a similar feature. In any case,
the kernel is named /dynix, so you can Configure it automaticly.
-- 
		[rbj@uunet 1] stty sane
		unknown mode: sane

lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall) (02/14/91)

In article <122475@uunet.UU.NET> rbj@uunet.UU.NET (Root Boy Jim) writes:
: It may be, but then maybe not. I have serious problems with
: TIOCSETP on a Sequent S81 Symmetry running Dynix 3.0.17.9.
: I sometimes get "improper operation for device" (or similar),
: or I get a floating point exception on TC's tcbreak script
: in the h2pl directory. Both of these work on Pyramid and Sun.
: I get the same results whether or not I use Sequent's original
: or suped up compiler (ATSCC), with or without -O or -g.
: 
: Larry, would you like a temporary account to fix this?

What, you want me to patch your OS?  :-)

: Also, Sequent has a parallel make facility. If you have a line like
: 
: 	target: & dependents
: 
: then dependents will be made in parallel. Quite handy when you
: have twenty processors at your disposal. Sequent's make seems
: to have macro '&' defined to '&' so using '$&' should cover
: any system. I think Encore has a similar feature. In any case,
: the kernel is named /dynix, so you can Configure it automaticly.

Maybe someday.  Writing metaconfig units for every gonzo feature that
somebody comes up with under the influence of cough syrup isn't at the top
of the queue right now.

Larry

hakanson@ogicse.ogi.edu (Marion Hakanson) (02/14/91)

In article <11424@jpl-devvax.JPL.NASA.GOV> lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall) writes:
>In article <122475@uunet.UU.NET> rbj@uunet.UU.NET (Root Boy Jim) writes:
>:. . .
>: Also, Sequent has a parallel make facility. If you have a line like
>: 
>: 	target: & dependents
>: 
>: then dependents will be made in parallel. Quite handy when you
>:. . .
>Maybe someday.  Writing metaconfig units for every gonzo feature that
>somebody comes up with under the influence of cough syrup isn't at the top
>of the queue right now.

Good choice, Larry.  Especially since there's a perfectly good make
utility which figures out the parallelism all by itself, without
needing any gross things added to cause your makefiles to be
non-portable.  Run out to your nearest GNU dealer, and pick up
version 3.58 or so.  Just think of all the time you'll save over
editing every makefile that comes across the net.  Now, where's
that cough syrup....

-- 
Marion Hakanson         Domain: hakanson@cse.ogi.edu
                        UUCP  : {hp-pcd,tektronix}!ogicse!hakanson

rbj@uunet.UU.NET (Root Boy Jim) (02/22/91)

In article <11424@jpl-devvax.JPL.NASA.GOV> lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall) writes:
?In article <122475@uunet.UU.NET> rbj@uunet.UU.NET (Root Boy Jim) writes:
?: It may be, but then maybe not. I have serious problems with
?: TIOCSETP on a Sequent S81 Symmetry running Dynix 3.0.17.9.
?: 
?: Larry, would you like a temporary account to fix this?
?
?What, you want me to patch your OS?  :-)

No, I want you to patch our compiler :-)
Actually, I tracked this down and fixed it myself.
For some reason, our machine doesn't cast negative floats
to unsigned correctly, and your program doesn't detect it.

You might want to undef CASTNEGFLOAT if you find /dynix.
I just edit config.sh by hand, as that's easier to do.

?: Also, Sequent has a parallel make facility...
?
?Maybe someday.  Writing metaconfig units for every gonzo feature that
?somebody comes up with under the influence of cough syrup isn't at the top
?of the queue right now.

Aw cmon! Parallel makes are a great idea.
-- 
		[rbj@uunet 1] stty sane
		unknown mode: sane

lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall) (02/22/91)

In article <123539@uunet.UU.NET> rbj@uunet.UU.NET (Root Boy Jim) writes:
: ?: Also, Sequent has a parallel make facility...
: ?
: ?Maybe someday.  Writing metaconfig units for every gonzo feature that
: ?somebody comes up with under the influence of cough syrup isn't at the top
: ?of the queue right now.
: 
: Aw cmon! Parallel makes are a great idea.

The "feature" I was complaining about wasn't the parallel make, but the
incompatible implementation thereof.

Larry