Re: gnubol: procedure types

[Tim Josling <TIMJOSLING@prodigy.net>]

As you point out, the preprocessor needs the text. It aso has to do multiple
(logical) passes of the data for example to do the replace verb, and nested copy
(an extension). The logical passes could be one physical pass but at present they
are multiple physical passes - it is simler that way. And it does 1,000 lines in
one meg. So I am not thinking of changing that.

In this context I would simply point out that there is no other open source
implementation of the cobol copy and replace that is complete and tested. Anyone
who wishes to put forward a more frugal piece of code is welcome. Note - *code*,
not design, concepts, emails etc.

I agree with the notion of having and API into the SYMT, so it can do memory
management among other things if needed. I will do that. But I will not be coding
memory management personally. That is what virtual storage is for.

At the same time I am not into egregious waste. As with GCC, you can clean up a
lot of stuff once it is used. Like, I have a handle under which I allocate all
the dispensible storage for a given x. Then at the end I do a 'free all memory
with handle x". It is not worth doing more than that IMHO, especially if it
starts creating bugs.

My target machine I have in mind is a single user machine running Linux. If he
wants to do a kernel recompile and rebuild the accounting application at the same
time he might get thrashing. So "don't do that".

Tim Josling




--
This message was sent through the gnu-cobol mailing list.  To remove yourself
from this mailing list, send a message to majordomo@lusars.net with the
words "unsubscribe gnu-cobol" in the message body.  For more information on
the GNU COBOL project, send mail to gnu-cobol-owner@lusars.net.