unit/procedure same name, unhelpful error

Waldek Hebisch hebisch at math.uni.wroc.pl
Wed Jul 11 01:33:54 CEST 2012

Jay Michael wrote:
> gpc 20070904, based on gcc-3.4.5 (mingw special)
>      I got a compilation error, "expression used as a statement".  When
> I finally realized I had a unit and a procedure in the unit with the
> same name, I changed one name and the error went away, so I'm guessing
> the duplication of name was somehow responsible.
>      I am not asking to be allowed to name a unit the same as some
> entity within the unit.

Actually currently you can (you probably should get an error
earlier).  Just write:


>  I am asking that the error message be changed
> to something that would have spared me thrashing around until I noticed
> that I had duplicated a name.
>      Failing that, could someone explain how what I did was an
> "expression used as a statement"?

I can explain how compiler produced this message.  Parsing


line compiler realised that this must be a procedure call,
so it attempted to call 'jsm2m'.  But it found that 'jsm2m'
names interface, not a procedure.  More precisely,
compiler was unable to find procedure or equivalent (gpc
allows some things which strictly speaking are not procedures
to be called like procedures) of name 'jsm2m', but 'jsm2m'
looked as something sensible (attempt at compiling it
did not produce an error).  Given this compiler decided
to emit this message.

I agree that error message is confusing.  OTOH what compiler
should do.  One could argue that compiler should emit error
message at your 'uses' statement.  However, 'uses' is an
UCSD Pascal constuct, which we tried to keep compatible
with Turbo Pascal.  And Turbo Pascal tends to ignore
name clashes due to imports.  If you use Extended Pascal
'import' statement then you do get error during import
and that is clearer:

jsm2.pas:2: error: bad import
jsm2m.pas:3: error:  redeclaration of `Jsm2m'
jsm2.pas:2: error:  previous declaration

For the compiler 'jsm2m' is an expression which produces a
value: corresponding interface.  It behaves as a somewhat
restricted record: you can use dot notation to extract
components (exports of the interface).

In your case message like below would be more appropriate:

  error: attempting to call interface

However, there is a lot of things which may appear at this
place, so trying to describe the object that compiler has
requires some effort.  OTOH, if there are no name clashes
then usually it is clear why the object can not be called.

Also, compiler has more or less the same information
when you write:

2 + 2;


a = b + c;

Note that the third one was probably intended to be assignment
statement, but in Pascal '=' tests for equality.  In both
cases the user probably tried to write a statement, but
for some reason it is incomplete.  Some people expect
that in such cases compiler should ignore the value,
and the message makes clear that this is illegal in
Pascal (one needs to do something with the value to
get a statement).

To put it differently, there are two typical scenarios
which from compiler point of view look the same:

1) user attempts to call something which is not a routine.
2) user wanted to write a statement, but it is somewhat
   incomplete and we only have part of it (that is
   an expression).

                              Waldek Hebisch
hebisch at math.uni.wroc.pl 

More information about the Gpc mailing list