On Jun 21, 2011, at 10:59 AM, Sven Van Caekenberghe wrote:
Hi Pavel,
On 21 Jun 2011, at 09:56, Pavel Krivanek wrote:
Hi,
I would like to point out to one new Pharo feature that is a potential issue. The package Zinc-Patch-HTTPSocket stole the ownership of several methods from the class HTTPSocket.
That is correct. It is a consequence of including Zinc as standard component in Pharo Core.
Note that many people think that most of HTTPSocket is of low quality and in need of being replaced.
Yes we should do that in 1.4. Any code to go in that direction are welcome.
That means that now this package doesn't act as an overlay patch.
I don't understand the exact meaning of 'overlay patch'.
The consequences are not fatal. It only means that now the Gofer package that uses this methods is dependent on Zinc and it makes a headless Gofer image a little bit larger and less clean.
Yes, currently the dependencies are Gofer -> HTTPSocket -> Zinc, but is could just as well (and better) be Gofer -> Zinc.
Zinc is definitively headless ready and I would hope clean enough.
I know that you want to reduce the code base as much as possible, but networking, http access and code loading can never really be removed I guess.
If you have remarks regarding Zinc itself, I would love to help making it as modular and kernel ready as possible.
Regards,
Sven