Xref: utzoo alt.msdos.programmer:2450 comp.sys.ibm.pc.misc:7536 comp.lang.c:37166
Path: utzoo!utgpu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!sdd.hp.com!news.cs.indiana.edu!cs.widener.edu!netnews.upenn.edu!gradient.cis.upenn.edu!jrv
From: jrv@gradient.cis.upenn.edu (JR VanMechelen)
Newsgroups: alt.msdos.programmer,comp.sys.ibm.pc.misc,comp.lang.c
Subject: Turbo C, a flex file, and a (nearly full) data segment (Uh-oh)
Message-ID: <39244@netnews.upenn.edu>
Date: 17 Mar 91 17:29:26 GMT
Sender: news@netnews.upenn.edu
Reply-To: jrv@gradient.cis.upenn.edu (JR VanMechelen)
Followup-To: alt.msdos.programmer
Organization: University of Pennsylvania
Lines: 23
Nntp-Posting-Host: gradient.cis.upenn.edu

I have a large flex specification, which, using flex 2.3, produces
large arrays of data.  I am using Turbo C v1.01.  Currently when I
compile the flex output, it uses 62548 bytes of the data segment. 
This is close to the 64K limit of a segment, and I assume that when I
reach that limit, I will not be able to proceed without extraordinary
measures.

However, the data in the output from flex is not in one single data
structure.  Microsoft C 4.0 has an option, "/Gt", which puts each
large data structure into a separate segment.  Is there a similar
option in Turbo C?  I couldn't find one in the manual.  Is there any
option other than to split the flex output manually?  Because I am
still modifying the specification, that method would be rather
unsavoury.  I already am generating the most compressed flex table,
but I really would prefer the fastest (and largest!) tables.

Thoughts, comments, flames?  e-mail, if possible, except flames.  I
prefer public humiliation :-)

Thanks,
JR

jrv@grad1.cis.upenn.edu