[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index][Thread Index][Top&Search][Original]

Re: [ID 20000125.002] Problem in documentation of Fcntl constants

From: Ilya Zakharevich <ilya@math.ohio-state.edu>
To: gsar@ActiveState.com (Gurusamy Sarathy)
Cc: barries@slaysys.com (Barrie Slaymaker), warren-odom@stenocall.com (Warren Odom), tchrist@chthon.perl.com (Tom Christiansen), perl5-porters@perl.org, gsar@ActiveState.com (Gurusamy Sarathy)
Date: Wed, 26 Jan 2000 15:43:22 -0500 (EST)
Message-Id: <200001262043.PAA06717@monk.mps.ohio-state.edu>

Gurusamy Sarathy writes:
> 
> On Tue, 25 Jan 2000 17:45:11 EST, Barrie Slaymaker wrote:
> >As Tom mentions later, it seems that it shouldn't be necessary.  And if
> >it is, it could be done with lower CPU overhead in atleast two ways:
> >just manually listing all of the prototypes, and by overloading import
> >and only eval()-ing when necessary.
> 
> I think the better approach is to make h2xs generate newCONSTSUB() entries
> for constants, and do away with encouraging the AUTOLOAD-based mechanism
> altogether.  ext/B/defsubs_h.PL does this already.

This in an enormous overhead.  Imagine loading a module with 6000
constants taking 6M just-for-nothing.

Putting prototypes may be a better solution for largish modules:
prototypes takes circa 120bytes per a proto.

Ilya

Follow-Ups from:: Gurusamy Sarathy <gsar@ActiveState.com>

References to:: Gurusamy Sarathy <gsar@ActiveState.com>

[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index][Thread Index][Top&Search][Original]