[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]
Re: crutch installation
- To: <roger>, <roland>, <xtech>
- Subject: Re: crutch installation
- From: Michael McClary <michael>
- Date: Sat, 17 Feb 90 18:29:15 PST
> From roger Sat Feb 17 17:12:11 1990
> 1) requires me to [know] what crutch does, and what the compiler doesn't do.
> this is such a can of worms that I'ld gladly pay a factor of 2 to not
> have to know the current state of disrepair of ATT's product. Also it is
> eaisy to overlook a construct that needs the crutch.
Not really. All you need to know is that if it produces one of a small
set of errors, changing its sourcefile name once will fix it forever.
On the other hand, if you don't want to be bothered, you can just name
everything YOU write or maintain with the name that runs it through crutch,
whether it's needed or not. In fact, if it will make your job easier, we
can make that "cxx", and use "cmm" for things that DON'T need crutch.
*I* don't want to twiddle my thumbs while every module I'm compiling
goes through an extra stage of processing it doesn't need.
> 2) the file name space doesn't need the additional confusion.
Agreed. Unfortunately, the alternatives are to run everything through
an extra preprocessor step, even if it's only to check that the extra
step isn't needed, or to put the confusion somewhere else (like the
makefile). Since c++ and c-- (alias c++ 2.0) are distinct languages
(that look a LOT alike), the filename is the appropriate place to
encode the distinction.
I agree that both approaches are not nice. I happen to want the
fastest compilitation I can get, and am willing to accept yet another
filename distinction to get it.
Let's do an experiment: Integrate the crutch with gpp via a pipe,
plug that into the gnatCC2 script, and see if the result is not
significantly slower. If this is true (and I suspect it may be),
I'll be OVERJOYED to drop my objection.