Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!ames!pasteur!agate!ucbvax!decwrl!sun!pitstop!sundc!seismo!uunet!portal!cup.portal.com!dan-hankins
From: dan-hankins@cup.portal.com (Daniel B Hankins)
Newsgroups: comp.sys.amiga
Subject: Re: Need recommendation on Modula-2 compiler
Message-ID: <13030@cup.portal.com>
Date: 29 Dec 88 06:38:23 GMT
References: <2058@van-bc.UUCP> <403@nth.UUCP>
Organization: The Portal System (TM)
Lines: 59

In article <403@nth.UUCP> loyd@nth.UUCP (Loyd Blankenship) writes a
scathing criticism of TDI's compiler.

     Personally, I don't think that it's that bad.  I've successfully
written a modestly sized program in it without too many problems.  I'm now
in the process of writing an interpreter for an object-oriented language in
it (not a small project).  Here's some answers to his points:

>1) The editor eats 50K of RAM every time it is invoked.  The only way to
>   recover the RAM is rebooting.

     Their editor is lousy anyway.  Don't use it.  MG or DME or just about
any other editor is superior.  MG in particular has the capability to be
customized to be syntax-sensitive.

>2) The editor sometimes finds errors, and sometimes doesn't.  Completely
>   random in functionality.

     Again, don't use TDI's editor.  Use the public domain program m2error
(available on Fish) which will invariably give the correct error code and
location.

>3) The example code given (the BOX code specifically) does not compile. 
>   Debugging is left as an excercise to the purchaser.

     I won't argue with you here.  I don't know, myself.  In my book,
compiling example code is usually a waste of time, unless it does something
in particular that you need.

>4) In the definition files, several variables and types are defined wrong.
>   It's been long enough that I don't remember specific examples.

     Again, no argument.  I haven't personally encountered any of these.  
Is this in the documentation, or on the files actually provided on the
disks?  I haven't yet encountered any errors in the files provided on the
disks.  If the definition file is wrong, well then I would just use DecSym
to generate one from the .sym file.  Haven't had a need to try that yet, so
it may not work.

>5) The documentation is the worst collection of vauge and misleading
>   statements it's ever been my misfortune to wade through.

     I must have backlevel documentation.  Mine is fairly short, clear, and
concise.

>6) The index is wrong about 25% of the time.

     You mean you don't grep the .def files on the distribution disks?

>7) Customer support is non-existant.  When you call on the phone (I tried
>   both the UK & the US phone #'s), the person who knows anything about it
>   is 'not available', and not *once* in over 20 attempts did anyone ever 
>   return my call.

     I haven't needed this yet, so I wouldn't know.  I get most of my
support from other users, when necessary.


Dan Hankins