TEAM-ADA Archives

Team Ada: Ada Programming Language Advocacy

TEAM-ADA@LISTSERV.ACM.ORG

Options: Use Classic View

Use Monospaced Font
Show Text Part by Default
Condense Mail Headers

Topic: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Date: Thu, 14 Nov 1996 15:09:56 -0700
Sender: "Team Ada: Ada Advocacy Issues (83 & 95)" <[log in to unmask]>
From: "GEMINI::RWATSON"@POLLUX.DASD.HONEYWELL.COM
Reply-To: "GEMINI::RWATSON"@POLLUX.DASD.HONEYWELL.COM
Parts/Attachments: text/plain (23 lines)
[log in to unmask]
X-Vmsmail-To: GEMINI::SMTP%"[log in to unmask]"
Message-Id: <[log in to unmask]>
Subject: TARTAN 5.1 Bug

I'm on a project that is using the TARTAN toolset, and we tried to use
their latest 5.1 compiler.  Unfortuanately, what would compile under their
old compiler now doesn't.  For those of you out there using the 5.1 beware
of using complicated addresses.  We had something like
DUMB_ARRAY(DUMB_INDEX)(TEMP_DUMB_INDEX_TYPE).DUMB_FLAGS
We load some values into a couple of registers for index registers off
of the base address of DUMB_ARRAY.  Unfortunately when we do the address
addition, we lose the base address part.  So instead of accessing
DUMB_ARRAY + r9+4028+(r10*1) we get r9+4028(r10*1)
You would think that with such an apparent bug TARTAN would want to get it
fixed right away. - WRONG.  Even with a support contract they said NO.
Yes we could try to play some games with this, but it would cost in
throughput.  It is kind of sad that a compiler company won't fix such an
apparent and basic error such as this.

Rick Watson
All opinions stated are my own, the company can't take credit for any of them.

ATOM RSS1 RSS2