[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: #if versus #ifdef in bison.simple
From: |
Akim Demaille |
Subject: |
Re: #if versus #ifdef in bison.simple |
Date: |
05 Dec 2001 09:35:52 +0100 |
User-agent: |
Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.4 (Civil Service) |
>>>>> "Paul" == Paul Eggert <address@hidden> writes:
>> That's true for bison 1.30. For the trunk, CPP macros in
>> bison.simple become more and more stricly internal. The managing of
>> the values of these macros will be only made by Bison by options in
>> the .y file.
Paul> But doesn't this introduce some compatibility issues unless
Paul> those macros are kept the way that they are, to support programs
Paul> that use the current interface?
Yes it does. We intend to provide backward compatibility, but to have
the documentation recommendations etc. promote the %directives instead
of #define.
Paul> YYERROR_VERBOSE itself may not be too much of a problem. I know
Paul> of only one GNU program that #defines it, namely a2ps, and a2ps
Paul> #defines it to 1.
Good news :)
>> Moreover, the documentation must be updated on the trunk!
Paul> And this is where my qualms come in. Once we are talking about
Paul> changing the documented interface, we have to be careful to
Paul> avoid gratuitous changes, while of course not standing in the
Paul> way of progress. (I know, I know, I'm sounding like a
Paul> politician. :-)
:)
We agree. One of our long term goals is to be able to address other
languages than C/C++. Therefore tuning the output via #define is
wrong, we have to move towards something more Bison.
But old .y in C should have the backward compatibility they need.