[Python-Dev] buildin vs. shared modules

[Martin v. Löwis]

"David LeBlanc" <whisper at oz.net> writes:

> What's the cost of mapping the world (all those entry points) at startup?

I believe it is measurable. It also adds maintenance costs to have
extension modules, both in terms of the build procedure, and in
packaging.

> You have to rebuild all of the main dll just to do something to one
> component. To me, that's maybe the biggest single issue.

When did you last wish to rebuild one of the modules without having a
PCBuild directory in the first place? If that ever happened, which
module did you wish to rebuild and why?

> Any possiblity of new bugs?

Not likely.

> Are app users/programmers going to have a bloat perception?

This is possible; it appears that all readers who, in this thread,
have spoken in favour of keeping the status quo have done so because
of a bloat perception.

> IMO, it contradicts the unix way of smaller, compartmentalized is better.

I dislike the usage of shared libraries on Unix, and still hope that
the Python build procedure becomes sane again by reducing its usage of
shared extension modules, in favour of a single complete binary.

> It's not unix we're talking about, but it still makes sense to me, whatever
> the OS.

It makes no sense to me whatsoever.

> On a related side note: has anyone done any investigation to
> determine which few percentage of the extensions account for 99% of
> the dll loads?

Do you have any specific concerns beyond FUD?

Regards,
Martin