Re: [RFC] Glib::filename_(to|from)_unicode => Glib->filename_(to|from)_unicode?

On Wednesday, December 24, 2003, at 10:05 AM, Ross McFarland wrote:

A. Pagaltzis said:
FWIW, I really hate the latter syntax used in cases where the
former would do.  At the same time, that syntax forces Perl to do
more work to find the function to call and pass the class name to
the function.  The only benefit of the arrow syntax is being able
to use inheritance, but is that really a concern here?

philosophically, the benefit of supporting inheritance everywhere is that you leave no spot where a calling layer *cannot* override something; i am not prescient enough to know ahead of time where some guy will find a need override filename_from_unicode to register all filenames that get converted, or add a croaking check to ensure that the drive is mounted, etc, etc.

I'd make
a case for all static "class methods" to be called with a package
qualifier rather than as class methods.

this is mainly a question of uniformity of api and if is it alright to break
something that's been put out there in a stable release.

indeed, uniformity of API is why i want to change it, and the fact that it will break things is why i'm putting it up for discussion.

while we're at it we probably ought to take out the prototype keyword
unless its in there for a specific reason as we haven't used it anywhere else.

the prototype tells the parser how many and what type arguments that particular function wants --- in this case, it wants one scalar, and the prototype keeps the parser from giving all the elements in a comma expression to the function call.

basically, the prototype allows you to omit the parentheses on calls to that function, making it behave more like a perl builtin.

muppet <scott at asofyet dot org>

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]