|
From: Theodore W. L. <tw...@sa...> - 2003-02-08 09:22:11
|
> -----Original Message----- > From: pyb...@li...=20 > [mailto:pyb...@li...] On=20 > Behalf Of Wari Wahab > Sent: Thursday, February 06, 2003 9:17 AM > To: will > Cc: Wari Wahab; Blake Winton; pyb...@li... > Subject: Re: [Pyblosxom-devel] Supporting comments, etc. >=20 >=20 > will wrote: >=20 > >The disadvantages to using XML would be that the performance of=20 > >loading/saving is probably an order of magnitude slower than=20 > pickling. > >In addition to that, we also take on the burden of=20 > architecting the XML > >format we want to save in and how it works. We definitely=20 > get more bang=20 > >for our buck with pickling. > > =20 > > > Sorry, the XML we were discussing about only relates to importing and=20 > exporting data between storageAPI compliant objects. If I don't like,=20 > say a file based storageAPI system, I can export to the=20 > pickle, and get=20 > the sqlite storageAPI object to get the data from the pickle=20 > export and=20 > store data in it's database, this is so that you don't lose=20 > data while=20 > changing 'drivers'. >=20 > >The advantage to using XML is that other non-Python processes can > >manipulate the data. I can't really think of other=20 > advantages--though it=20 > >should be noted that this is somewhat of a big one. > > =20 > > > Since this is a dump() and load() methods, maybe later we can=20 > define an=20 > export() method to export to a documented format like XML or maybe=20 > MovableType style comments data. A dump() and load() is for pyblosxom=20 > and pyblosxom only I guess (though other program could use its data) >=20 > >How does blosxom and the other *osxom systems store data? Would we=20 > >want > >to store our data in a compatible format further enabling=20 > folks to switch=20 > >between *osxoms? > > > There's no common way, Rael's blosxom uses a modified standalone=20 > trackback tool for comments (which uses perl style dumps),=20 > I'm not sure=20 > what Sam Ruby uses, phosxom uses sqlite to store feedback.=20 > For pyblosxom=20 > I want it as flexible as possible yet can migrate data as well. Data=20 > storage is one thing, exporting it is another. >=20 Sam is using text files written into directories. They are not XML. There's a lot of value to using XML as an interchange format. In fact, = when I saw Sam 2 weeks ago (almost) we talked about using RSS as a = representation for articles as well as comments. Of course, we were also talking about wrapping the RSS item in a SOAP envelope for the next generation of the weblog / Metaweblog API -- that's an aside. I haven't played much with Python's XML support, but I know that at = least one of the python XML libraries is based on expat which is a very fast C parser. I'd like to try the XML format and see if performance is really a = problem. This might mean using more libraries then what's in the Python base, = though. Ted=20 |