Improvement request
lazarus / 64-FPC development

----------------------------------------------------------------------
440bx · 2024-02-10 14:06

Hello,

FPC v3.2.2 currently emits an "Error: Illegal expression" message when
it encounters an assignment statement with a missing ":", i.e, Code:
Pascal  [Select][+][-]window.onload = function(){var x1 =
document.getElementById("main_content_section"); if (x1) { var x =
document.getElementsByClassName("geshi");for (var i = 0; i SomeVar =
avalue;   { illegal expression - note missing ":" }
This is an easy mistake to make when porting C code to Pascal.  More
often than not, the cause of the problem is evident (the missing ":")
but, on occasion, after porting many lines of code, the reason can be
overlooked resulting in a waste of time.

The request is therefore: if at all possible improve the error message
to something along the lines of "possible ":" missing in assignment
statement" or something along those lines.

Thank you.

----------------------------------------------------------------------
Kays · 2024-02-10 22:40

The word note is in the imperative mood. “Pay attention!” Maybe you
can’t see it but there ain’t no colon.

Your wording on the other hand is pretty suggestive. We don’t know
what’s missing; there are multiple valid hypotheses.

----------------------------------------------------------------------
440bx · 2024-02-10 23:41

Quote from: Kays on February 10, 2024, 11:40:02 pmthere are multiple
valid hypotheses.
My hypothesis is that you should follow Lincoln's advice.  It would
serve you well.

----------------------------------------------------------------------
Leledumbo · 2024-02-15 15:40

Quote from: 440bx on February 10, 2024, 03:06:55 pmThe request is
therefore: if at all possible improve the error message to something
along the lines of "possible ":" missing in assignment statement" or
something along those lines.
In theory, it should be possible, as an explicitly handled case, like
this one where it specifically handles proc():=. In this case, instead
of falling into an else clause of a case statement, a case to handle
_EQ should be added.

----------------------------------------------------------------------
Thaddy · 2024-02-15 17:34

That is not an improvement, it is a fallacy.
"Fixing" mistakes by changing syntax is a mistake.
Not to mention that assignment does not  equal equality.
The same is the case in C , = vs ==. so shut up is the most kind
expression I could find.

----------------------------------------------------------------------
xixixi · 2024-02-15 19:33

I think it's a great improvement. If the statement is an (illegal)
equality expression, a note could be added to "Illegal expression",
like "Did you mean an assignment ':='?".

----------------------------------------------------------------------
440bx · 2024-02-15 20:35

Quote from: Thaddy on February 15, 2024, 06:34:39 pmThat is not an
improvement, it is a fallacy.
You and Kay should get a room... make each other happy <chuckle>

----------------------------------------------------------------------
Thaddy · 2024-02-15 20:48

Quote from: 440bx on February 15, 2024, 09:35:52 pmQuote from: Thaddy
on February 15, 2024, 06:34:39 pmThat is not an improvement, it is a
fallacy.
You and Kay should get a room... make each other happy <chuckle>
And you should be able to distinguish between assignment and
comparison.

----------------------------------------------------------------------
dbannon · 2024-02-16 02:09

Oh, come on, 440bx is not suggesting the compiler should go off and
edit your source or anything like that. Just issue a more meaningful
message. One that would be more helpful in the vast majority of cases.

Davo