Re: [Dbmail-dev] minor fixes + some small cleanups

[Dan Weber]

Ilja Booij wrote:

Dan Weber wrote:

I guess thunderbird doesn't like copy and paste.

2004-07-01  Dan Weber <[EMAIL PROTECTED]>

       * auth/authsql.c, db.c: replaced function definitions with 0
       arguments with a void argument.

Is this necessary or just good practice? If so, we can apply this later on.

Necessary. The C compiler doesn't do type checking so you should fill an empty argumented function with void. Otherwise they can just pass unlimited arguments to the function.

       * db.h, mysql/dbmysql.c, pgsql/dbpgsql.c (db_num_rows):
    changed
       type cast from unsigned to long long as suggested by Intel
       Compiler.

Why should this be a long long? I can't see the reason for it. What's the output from the Intel compiler?

The Intel Compiler claimed that res was a long long and was being converted to an unsigned on return. It said you were probably losing bytes.

       * None: Made compatibility for more compilers, most notably
    tcc
       and icc.

Which changes do that?

Above changes. ICC is much stricter and has a much better lint than gcc, its worth a try. For non-commercial its free as in beer.

       * Makefile.am, auth/Makefile.am, mysql/Makefile.am
    (AM_CFLAGS):
       updated to pass a preprocessor definition at compile time if
    using
       mysql.

       * auth/authsql.c, db.c, db.h (db_insert_result): placed ifdef
       statements around places where db_insert_result is being
    called in
       an effort to not pass an unused argument to dbmysql.c's
       db_insert_result.

       * mysql/dbmysql.c (db_insert_result): replaced the UNUSED
    argument
       for db_insert_result with void, followed by changes with
       preprocessor to make compatibility through the source.  For
       changes with preprocessor see above.

Other compiler probably cannot handle the UNUSED thingie. I hate to clutter up the sources with those #ifdef's though, as they make the code less clear & clean.

Sure. You are right, most other compilers don't use the UNUSED thing. I am sure you can make a preprocessor statement thats much smaller and concise to double preprocess.

To conclude:





Ilja

Compatibility with other compilers is not so high on my list that I'm going to make changes to the current release candidate for it. I'll only make changes that are absolutely necessary. I would've like to tell you to put these patches into a bugtracker entry, but the dbmail.org and it's bugtracker are currently off-line, so that's not possible, I'm afraid. (BTW: We're working on it :) )

Dan Weber

signature.asc
Description: OpenPGP digital signature