
bernhard at intevation
Mar 2, 2009, 11:32 PM
Post #18 of 18
(1658 views)
Permalink
|
Igor, James, thanks for your responses and for doing nice python packages of gpgme as Free Software! I suggest you should both use the 1.0 label for the next release. Think about that release numbers are mainly for journalists. ;) Am Montag, 2. März 2009 18:31:21 schrieb James Henstridge: > On Mon, Mar 2, 2009 at 11:23 AM, Igor Belyi <belyi [at] users> wrote: > > Yes, I think this is the right assessment of the implementation > > difference between pyme and pygpgme - pyme uses SWIG to have python > > binding for gpgme functions and then uses extra python classes/packages > > for pythonizing GPGME interface where as pygpgme does most of the work > > using Pyhton C API directly. In short, pyme hides behind SWIG from > > handling difference in Python and GPGME versions, where as pygpgme takes > > approach of having less dependencies on another product. > > The choice not to use swig was not an issue of dependencies, but > rather bad experience with it in the past. We (at Intevation) also found that swig was a good idea in 2000 and used it for pyshapelib. Later we found it was more getting in the way, so we probably will not use it for the next python wrapper we do. And pyshapelib has meanwhile be ported to be native. > I've seen too many memory > leaks or crasher bugs in swig based bindings to trust it. I needed a > binding that could be used by a long running web service without > causing it to run out of memory. So far I did not try pygpgme yet, but pyme which worked fine for a couple of short running scripts. I know that roundup 1.4 uses pyme and roundup often is run as a long term server process. This is a good sign for pyme. Best, Bernhard -- Managing Director - Owner: www.intevation.net (Free Software Company) Germany Coordinator: fsfeurope.org. Coordinator: www.Kolab-Konsortium.com. Intevation GmbH, Osnabrück, DE; Amtsgericht Osnabrück, HRB 18998 Geschäftsführer Frank Koormann, Bernhard Reiter, Dr. Jan-Oliver Wagner
|