[Tcl-bugs] [ tcl-Bugs-1029267 ] bad [expr] syntax error message

[SourceForge.net <no...@so...>]

Bugs item #1029267, was opened at 2004-09-16 19:22
Message generated for change (Comment added) made by flatworm
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110894&aid=1029267&group_id=10894

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: 44. Parsing and Eval
Group: obsolete: 8.5a2
Status: Open
Resolution: None
Priority: 7
Submitted By: Don Porter (dgp)
Assigned to: Don Porter (dgp)
Summary: bad [expr] syntax error message

Initial Comment:
% expr 1=2 
syntax error in expression "1=2": extra tokens at end
of expression


The error message
"single equality character not legal in expressions"
is present in tclParseExpr.c and would
be more fitting here.  Why isn't it returned?

----------------------------------------------------------------------

Comment By: Konstantin Khomoutov (flatworm)
Date: 2006-06-01 21:06

Message:
Logged In: YES 
user_id=1350198

A kind of "silent error" when just one double quote is
passed to the expr:

% if {"} { puts then } else { puts else }

% set errorCode
NONE
% set errorInfo

    while executing
"if {"} { puts then } else { puts else }"

On the other hand,
if {(")} ...
produces the sensible error message:
missing "

----------------------------------------------------------------------

Comment By: Donal K. Fellows (dkf)
Date: 2004-09-19 21:11

Message:
Logged In: YES 
user_id=79902

Getting sensible failures out of a parser is notoriously
difficult. At least we aren't in the previous situation
where all the parser would say was "syntax error", which was
singularly uninformative in complex expressions...

----------------------------------------------------------------------

Comment By: miguel sofer (msofer)
Date: 2004-09-19 19:20

Message:
Logged In: YES 
user_id=148712

Another one:
% expr cos(1=2)
syntax error in expression "cos(1=2)": missing close
parenthesis at end of function call

What is happening is that the expr parser considers the
subexpression '1=2' finished at the '1', with extra tokens
'=2)'. 

----------------------------------------------------------------------

Comment By: miguel sofer (msofer)
Date: 2004-09-19 18:24

Message:
Logged In: YES 
user_id=148712

At least some of the tests (notably parseExpr-16.33) expect
this behaviour.

----------------------------------------------------------------------

Comment By: miguel sofer (msofer)
Date: 2004-09-19 18:18

Message:
Logged In: YES 
user_id=148712

Oh my ... the whole "wrong lexeme type" section of
ParsePrimaryExpr() seems non-functional. Another example is
 % expr 1,2
 syntax error in expression "1,2": extra tokens at end of
expression

What is happening is that the parser sees the '1' as a
complete literal expression, and complains about '=2' or
',2' or 'a2' following it. The correct message is produced
for [expr =2], as the first lexeme is the bare = sign.

Another beauty:
% expr (1=2)
syntax error in expression "(1=2)": looking for close
parenthesis



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110894&aid=1029267&group_id=10894