NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/os/vms/12773

Internet Message Format | 1992-07-25 | 1.3 KB

Path: sparky!uunet!cis.ohio-state.edu!pacific.mps.ohio-state.edu!linac!att!ucbvax!galaxy.dnet!gleeve
From: [e-mail]
Newsgroups: comp.os.vms
Subject: vms mem mgt
Message-ID: <[e-mail]>
Date: 24 Jul 92 15:47:04 GMT
Sender: [e-mail]
Distribution: world
Organization: The Internet
Lines: 18

In the coding of AnalytiCalc I also had to deal with a large but
ordinarily sparse structure. I arranged all accesses to the structure
to be done via calls to a couple routines (a put and a get), then
could modify the routines to access a more limited amount of space
sensibly using files, hash tables, trees, or what have you. If the
person who has to handle his quarter-gig of space can arrange it,
that's a good approach, and it does not need to mess with allocating
lots of tiny parts of process address space.
   It seems conceivable to me that one could insert an exception handler
to catch access violations, then have a routine trap these exceptions
and access a more reasonable structure as I did (and everyone who writes
a spreadsheet does, almost). Just lie to the caller about the address of
the structure so ALL the accesses he tries will trap. Even the overhead
of a condition handler is probably less than that of all the $cretva
calls, and the structure, more sensible.
Glenn
[e-mail]