[omniORB] omniidl2 compiler bug ????? in omniORB2.6.0

[Gary D. Duzan]

In Message <3olnmbj5uy.fsf@neem.cam-orl.co.uk> ,
   Sai-Lai Lo <S.Lo@orl.co.uk> wrote:

=>Yes, this bug in omniidl2 affects both HPUX and AIX port. AFAIK the bug
=>does not affect other platforms because the preprocessor on the other platforms
=>do not generate the offending #line <line no> directive.

   Now, to be fair, it isn't so much that the #line directives were
"offending". What was really offending was that the code that deals
with them was committing a cardinal sin (in C terms) by reading past
the end of a string. No sense blaming poor, helpless #line directives
who are only doing their job. :-)

					Gary Duzan
					GTE Laboratories



=>I'll be releasing 2.6.1 shortly. The bug is fixed in that release.
=>Stay tuned.
=>
=>Sai-Lai
=>
=>
=>>>>>> Gary D Duzan writes:
=>
=>>    I haven't checked your IDL specifically, but it sounds somewhat like
=>> a bug I uncovered recently which had to do with proprocessor #line
=>> directives. Try the attached patch and see if it helps. You'll probably
=>> have to do a:
=>
=>> =>
=>> =>it's me again and I have found a strange behaviour of the idl compiler
=>> =>in omniORB 2.6.0.
=>> =>It was possible to reduce my very complex idl files to an example which
=>> =>shows following
=>> =>error while compile time:
=>> =>
=>> =>omniidl2: "/tmp/idli_a05314.cc", line 16: error in lookup of symbol:
=>> =>MyException
=>> =>omniidl2: Main.idl: found 1 error
=>> =>
=>> =>This error indicates that the include statement doesn't work. Some
=>> =>changes to the code
=>> =>leave to a complete compilation (all changes to Main.idl)
=>> =>
=>> =>If I delete the comment in the first line -> it works
=>> =>If I delete the comment only, leaving two lines before the #ifndef -> it
=>> =>doesn't work
=>> =>If I insert a comment line "// abcdefg" between #define and first
=>> =>#include -> it works
=>> =>
=>> =>I think it's very very very strange. Does anybody be able to reconstruct
=>> =>this error with his
=>> =>2.6 idl compiler? If yes it could be a bug (one have to be careful with
=>> =>such a statement right?)
=>> =>If no, it could help us to find hints for the error source
=>> =>
=>> =>Best regards
=>> =>
=>> =>Ralf Walther
=>
=>-- 
=>Dr. Sai-Lai Lo                          |       Research Scientist
=>                                        |
=>E-mail:         S.Lo@orl.co.uk          |       Olivetti & Oracle Research Lab
=>                                        |       24a Trumpington Street
=>Tel:            +44 223 343000          |       Cambridge CB2 1QA
=>Fax:            +44 223 313542          |       ENGLAND
=>