home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.ada
- Path: sparky!uunet!milo!drew
- From: drew@verdix.com (Drew Johnson)
- Subject: Re: Need new idea to get around sunada 1.0 [Verdix] error
- Message-ID: <1992Aug27.130456.444@verdix.com>
- Organization: Verdix Corp.
- References: <1992Aug19.185609.10211@src.honeywell.com> <vera.714899225@fanaraaken.Stanford.EDU>
- Date: Thu, 27 Aug 1992 13:04:56 GMT
- Lines: 46
-
- In article <vera.714899225@fanaraaken.Stanford.EDU> vera@fanaraaken.Stanford.EDU (James S. Vera) writes:
- >Ack! The old Verdix bug. Our group first complained about that bug
- >half a DECADE ago. Bug progress has been made. Instead of barfing up
- >with an assertion error, Verdix is instead (in some future, unnammed
- >release) going to issue a standard error message. Big improvement.
-
- We have since decided to make some changes (but the error message will
- still get put in too).
-
- >
- >The problem is not the case statement, per se. Instead the problem is
- >that Verdix has a hard limit on the number of expressions which can
- >occur in a subprogram. The limit is something like 20,000
- >expressions.
- >
- >To get around this bug, you need to break the offending subprogram
- >into multiple subprograms. This, of course, is a pain in the but when
- >the subprogram is generated automagicly but gee, you don't expect your
- >Ada compiler to compile valid Ada code, do you?
-
- It comes down to "how big is too big?" We used to think that subprograms
- like this were "too big". We are revising this because it is apparent
- that this is not the case. Unfortunately, there is always going to be
- a limit to the size of a single body of source that the compiler will
- accept. What will this limit be? Well, this is difficult to translate
- into lines, because a single line can be very complex or very simple.
- Right now, I can only tell you that it will be substantially increased
- in the next version of the compiler (6.2 for most platforms, SunAda 2.0
- for Suns). Does this mean your code will now compile? I don't know, because
- I don't know just how far over the limit you were. You might still be
- over the new limit. This is doubtful, but possible.
-
- >I've pointed out this Verdix debility to the AYACC folk and they said
- >they'll look into it. I'd put money on it that the AYACC folk code
- >around Veridex's bug before Veridex fixes it.
-
- Well, we will have beta compilers in the field with this fix in about
- a month (less for some platforms). It would still be a good idea for
- AYACC to break up huge subprograms, because even though we will accept
- larger ones with 6.2, it is going to be a huge resource sink. I suspect
- that if you compiled this same code with other compilers, you will find
- the same thing.
-
- Drew Johnson, QA Manager
- Verdix
- drew@verdix.com
-