Older blog entries for edd (starting at number 51)

Keeping it brief: been very busy due to XML Europe work. However, managed to package four new upstream BlueZ releases for Debian. Also did more SyncML-related work. I had been going to junk my XML/WBXML translator in favour of wbxmllib, but found that while wbbxmllib is excellent, it wasn't suitable for my one, very specific purpose. So I'll press on with my work but won't bother generalising it too much, as those who want a general WBXML toolkit can't go far wrong with wbxmllib.

Finally got a few moments to do a bit more hacking on my phone manager applet. There's something rather satisfying about having it reconnect automatically to my Bluetooth-enabled phone when I wander back into the house. Anyway, I got message sending going now, too. The remaining task to make it really useful is hooking it up to one or more addressbook sources.

Finalised the picks for the Apps track of the O'Reilly Open Source Convention. We've got 14 talks, spanning desktop, hacking, sysadmin and standards. The quality of submissions was excellent, and I'm very pleased with the way the track's looking.

The needle almost went off the distract-o-meter scale this last week. Need to pull it back into focus for the coming week. Here's some highs and lows of my free software work.

Highs: bluez-pan got accepted into the Debian sid archive (enables dial-up and ethernet networking over bluetooth devices.) Also the debs of the Epiphany browser (called epiphany-browser in sid) got accepted. I am struggling to identify whether this is good, or I've just laid myself open to processing endless bug reports. Finally, did some work on gnome-bluetooth which separates out the bluetooth GObject from the bonobo object, I have a user who just wants the GObject stuff for use in a handheld environment. Again, users are good, but mean more work!

Lows: failed miserably to get Epiphany compiling against MiniMo, a reduced footprint branch of Mozilla. The idea is the for embedders, you ship your own small distribution of Mozilla along with your browser. This would be ideal for Debian, where the only other alternative is compiling against the mozilla-snapshot. Unfortunately mozilla-snapshot moves at a slower pace than Epiphany, which is tracking Mozilla's CVS HEAD. I guess I just don't have enough Mozilla-l33tness and, per my policy of not getting too distracted, really oughtn't to spend the time acquiring it just now. Sigh.

16 Feb 2003 (updated 16 Feb 2003 at 19:41 UTC) »

Another 11pm project that got out of hand.... Out of curiosity I tried compiling epiphany, a new Gecko-based browser for GNOME: it strips all the excess away and concentrates on being functional and clean. After half an hour I was utterly hooked and wanted Debian packages for my system. At 3.30am I had them, after a few hiccups.

Somebody else had filed the ITP for epiphany on Debian but not pursued it recently, we're now talking about co-maintaining it. So, with a bit of luck epiphany could hit Debian unstable soon. Now, back to all those other bits of work I really needed to do.

11 Feb 2003 (updated 12 Feb 2003 at 00:02 UTC) »

For various reasons I've found a namespace-aware XML parser for C++ hard to come by (no thanks, I don't want Xerces). So today I wrote a first attempt at a C++ wrapper for the XmlTextReader API from DV's libxml2. It's incomplete, but it works, so I sent it to the libxml++ mailing list. Unfortunately this list doesn't seem to be archived online anywhere. I expect to be berated for my poor understanding of C++ and lousy style etc, etc, but I hope showing willing will get some movement going on this.

Now I have a lightweight C++ namespace-aware XML parser I can carry on with my SyncML work... It's amazing how many diversions this project is taking me on.

Took some time today to familiarise myself with walters' build system for Debian packages, Colin's Build System. As a result, I'm happy to present an unofficial Debian repository for my GNOME/Bluetooth hackings. These packages aren't of a standard fit for the distribution yet, but it should make testing them a lot easier, as compiling from source means dragging in a lot of dependencies.

Spent quite a bit of time in the last week on Bluetooth work. First I uploaded the new BlueZ SDP libs to Debian sid, hopefully they'll make it in next week. Secondly, I packaged up the BlueZ PAN tools for sid, too, I'll upload them when the new SDP libs hit the archive. This means that Debian unstable users will soon easily be able to do things like allow PPP logins over Bluetooth devices.

I then updated my GNOME Bluetooth Subsystem to use the new SDP API, and finished off the primitive GUI I made for it while in Baltimore airport last year. It's all very basic and alpha, and in one way I'm depressed about releasing code so early-stage, but I know it'd never get out at all if I didn't!

31 Jan 2003 (updated 31 Jan 2003 at 10:39 UTC) »

Hooray, the WriteTheWeb relaunch is complete. Kudos to the Plone guys for a useful CMS product.

With regard to the "unhelpful comments" thread, I had a relevant experience today. My correspondent asked why Phone Manager was not in public CVS anywhere, why not hosted on SourceForge, why in C++ not in C, and why it didn't have a mailing list.

Seems that the minimum expectations of consumers of open source code are rising somewhat!

(For the record: I find SourceForge clunky and too-oft unavailable, and have my own resources anyway; a mailing list for such a pre-alpha bit of code would encourage an amount of correspondence I don't yet have time for; I have a private CVS; and I made it C++ because I hate you all!)

O'Reilly Open Source Conference

I'm lucky enough to be chairing the Applications Track at O'Reilly's OSCON this year. I've just put up the Call for Participation for this track. If you're interested please head over and consider submitting a talk proposal.

coda

I spent too much time yesterday trying to get a distributed file system running. I gave up quite quickly on InterMezzo because of its instability (at least on my system.) I then spent a good while on Coda, which was tantalizingly close to working before it failed. I'm grateful to Jan Harkes for spending a lot of time with me trying to track down the issue (filesystem failed to reintegrate after reconnection) but we didn't really manage to resolve anything last night. Sigh, looks like I'm stuck with NFS for the while.

SyncML

I finished the first stage in my SyncML implementation work, which was a converter from the WBXML encoding to XML -- I want only to cope with one encoding in the actual SyncML processor itself. The WBXML converter's written in C++, but I intend to do the SyncML manipulation in Python. (This is partly because I couldn't find a C++ XML parser that supported namespaces and had a GPL license, but also partly because I want the flexibility of a scripting language while I fool around with this project.) I've started a SyncML page, which has the first release of this software on it.

As I was starting to write this work up, I discovered a WBXML implementation that must have been started around the same time as mine, released on SourceForge: wbxmllib. If I'd seen this before I'd have been saved days of work, but sometimes things happen this way, I guess. I've now got to decide whether my WBXML parser has a long term future or not.

At the end of the day if nothing else, I figured out some automake for building and installing Python extensions with SWIG.

plone

Over the holiday break I worked on a relaunch of one of my existing web sites, using plone, a content management system based on Zope. I'm very impressed! Combined with my modest Zope skills, I've been able to get an impressive amount of work done in a short time.

The FTP/WebDAV features of Zope made importing the old content pretty easy. My legacy content was in a PHP/MySQL system: I just wrote another template which created barebones HTML with some information into metatags, spidered these with wget, and then copied them via WebDAV into plone.

42 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!