| Clark F. Morris, Jr. 2004-03-28, 9:50 pm |
| Rick Smith wrote:
> Lueko Willms <l.willms@jpberlin.de> wrote in message
> news:95jJFQJeflB@jpberlin-l.willms.jpberlin.de...
> [snip]
>
>
>
> It cannot have gone on too long because it has not been
> resolved. One more point needs to be raised, in context.
> One or two more exchanges should do it.
>
>
>
It would be very interesting if someone were to post a synopsis of the
debate to bit.listserv.techwr-l (if someone sends me a copy of one of
the more succinct exchanges I will post it to the list
techwr-l@lists.raycomm.com which mirrors to the bit.listerv list). Our
users sometimes complain about the cryptic error messages that we give
them. The question is not so much whether the error message in question
can technically be called a syntax error but rather does that message
clearly indicate where the person should look to correct the problem.
An error message should hopefully do more than just let you know that
something is wrong and this is the area in which it is wrong. It should
as clearly as possible let you know the specific problem. Over 20 years
ago, I enthusiastically upgraded MVS SP from version 1 release 3
modification 1 to version 1 release 3 modification 2 because that
release finally had the converter interpreter inform someone which
symbol was not defined in a procedure rather than giving the cryptic
"SYMBOL NOT DEFINED IN PROCEDURE". A procedure is a canned piece of JOB
CONTROL LANGUAGE this is invoked by an EXEC statement, JCL being a
primitive equivalent of the Burroughs WFL.
When we write programs, our error messages should give as much
indication as to what the problem is as possible and if they are
interactive, the help should be able to go into excruciating detail if
needed. All error messages should be judged by the criteria of giving
enough indication as to what the problem is and what is needed to fix
the problem. My reading of the discussion is that the message in
discussion did not. If in fact it said that the syntax error caused the
compiler to exceed limits or otherwise gave a clue as to the type of
syntax error, then it was adequate. If it was of the nature of the
generic "SYMBOL NOT DEFINED IN PROCEDURE" message where it didn't tell
you which symbol, then it did not.
|