Older blog entries for lloydwood (starting at number 52)

Behlendorf admitted that at least half of all the 129,000 open source projects on Sourceforge are dead--no one is working on them.

-- That word "open", Brad Grimes and Joab Jackson, 15 September 2006.

I think 'at least half' is a trifle misleading. I maintain SaVi. I'm the only maintainer and active developer; I do very little work on the program, it's of minority interest, received patches are rare, and I'm lucky to issue three minor revisions per year to keep it running and up-to-date with compilers and environments. It's about at the bare acceptable minimum level of maintenance and popularity for a Sourceforge project. But, despite my lack of work and the lack of interest and popularity, the SaVi statistics consistently show that SaVi is in the top 1% of active projects on Sourceforge.

Over 90% of the projects on Sourceforge are dead. Yes, that's at least half, but it's also so much more. Were Sourceforge a garden, a thousand flowers would be blooming - and a hundred thousand more would be dead and decomposing. It's a very long tail.

In other news, I've finished the user interface tweaks needed to make SaVi look good on Mac OS X. This is now available in the SaVi development snapshot.

It's been announced that Advogato is going offline. I do have my other blogs; I'll probably be posting intermittently elsewhere.

Released SaVi 1.3.1 today. SaVi now works on Mac OS X - the screenshot was good to see.

Now we have to figure out why Geomview crashes on Mac OS X only when launching SaVi; they work fine apart, but not together. The crash is on Intel Mac (when Geomview is built with gcc 4.0.1) and PowerPC Mac (with gcc 3.3 and gcc 4.0.1). Not getting along with the POSIX pipe handling is the best current bet. If you're a Mac programmer and you know better, please tell the geomview-users list.

Sourceforge had logins disabled today, and now the uploads directory is clogged with files and anonymous ftp is disallowed. Sometimes it's hard not to look a gift horse in the mouse.

After a few weeks of cheerleading and of testing packages on my part, we now have a test release candidate of Geomview that, well... actually works. Reliably.

Giving a new lease of life to an old package like this is good -- there are lots of mathematicians and students who could never fix the package, but will benefit from being able to use it. And since it's under LGPL and can now be built cleanly from Sourceforge CVS... hopefully more people will add fixes over time. (Hoffentlich. Such a nice German word.)

The major contributions by Claus were cleaning up the autoconf stuff, making it straightforward for anyone to regenerate working configure and makefiles, and fixing the flaky file- and pipe-handling. And now he's done the hard stuff, he's onto the easy low-hanging fruit -- porting old modules into the new autoconf framework, even considering Mac OS X compatibility.

My next release of SaVi will be nagging everyone to upgrade their copies of Geomview. It's more than worthwhile.

19 Jul 2006 (updated 19 Jul 2006 at 17:14 UTC) »

Last week, as a guest lecturer at the International Space University summer session in Strasbourg, I ran a three-hour workshop introducing students to satellite constellations. That workshop made heavy use of SaVi in the slides and exercises.

Despite my fears, SaVi itself worked smoothly. Geomview and Cygwin on the classroom's available Windows PCs were more problematic, and had to be quickly restarted a couple of times. (Geomview currently has pipe communication problems on non-glibc systems. Cygwin's X server is very fussy about file permissions when starting, and fails hard at any opportunity. The latter was a surprise.)

Setting everything up for the tight deadline of the workshop encouraged more work on SaVi -- some minor bugs fixed, new simulation scripts, and, finally, prompted a SaVi user manual so that the students would have something to read. But that's just minor work and the icing on the cake after a few years of development; the major work had already been done.

It's good to know that the software I've been maintaining is good enough to throw at new users and have them use it successfully in only a few minutes - and good to finally use my own project as part of the daily job! I've now released SaVi 1.3, including all the work done in France.

25 Apr 2006 (updated 27 Apr 2006 at 01:04 UTC) »

After a several-year hiatus, I've created a cafepress store with my ++ungood; and ++good; T-shirt designs. The designs are very simple, but programmers (including Guido van Rossum of Python fame) seem to enjoy the ++ungood; pun.

++ungood; black shirt ++ungood; white top


The designs were originally sold by Need to Know. NTK's shop remains in Airstrip One, on the wrong side of the pond from the proles in the bulk of Oceania, so parallel availability from cafepress, in the heart of the homeland itself, should redress this.

Without Dave Green and NTK this simple idea would not have come to fruition. Dave believed in my idea, and added to it. While I spent seven years of my life believing in and adding to the ideas of NTK, which, in its changed form, continues to barely exist as a mere shadow of its former incisive self. (As do I.)

Still, change is inevitable. As Wired used to say: change is good. They were, of course, wrong. (Well, small change is good. Large notes are better. Everybody sells out.)

Thanks, guys.

18 Mar 2006 (updated 1 Feb 2007 at 04:12 UTC) »

From The Economist:

Projects that fail to cope with open source's vulnerabilities usually fall by the wayside. Indeed, almost all of them meet this end. Of the roughly 130,000 open-source projects on SourceForge.net, an online hub for open-source software projects, only a few hundred are active, and fewer still will ever lead to a useful product. The most important thing holding back the open-source model, apparently, is itself.

-- Open, but not as usual, 16 March 2006.

In a recent email newsletter, Sourceforge admins said that clearing up dead projects was a much-requested feature, and that they would be working on doing so.

After grumbling once more about how behind the times Geomview development is, I've been made a project administrator. This means that I could create the next Geomview release, if only I could get to grips with autoconf, imake and the complex build system; I can't actually build from CVS head. More practically, it limits my complaining about things that I now theoretically have the power to correct myself. We really need to get Geomview 1.8.2 released; 1.8.1 requires substantial patches to compile with anything later than gcc 2.9.5, and is showing its age. 2001 is a long time ago as far as compilers are concerned.

Released SaVi 1.2.8, bringing its fixed fisheye to a wider audience. There's probably some more code cleanup I could do on the fisheye and elsewhere (including getting to grips with properly handling two-line element sets and time for simulating real satellites), but that will just have to wait for another Christmas holiday...

And, when I wasn't modifying SaVi and the family television was finally free, I spent five late nights getting to Pro-Master rank in SSX On Tour. It's a fine followup to SSX Tricky and SSX 3. Yes, the effort Electronic Arts exerts from its employees seems to be entirely worthwhile.

So, yet another Christmas spent with virtual snowboards and virtual satellites. Kids! Simulation is fun!

Did some more SaVi development, finally discarding the X-only fisheye plot code in favour of reworked code that uses the Tk canvas widget throughout instead. In the process, the executable shrank by about 20%, too; I don't think the X libraries will be missed.

By not requiring X, SaVi is finally properly useful on Cygwin and with its weird Insight Tcl. So proper Mac OS X support can't be too far away, either, surely? But given that I've been using Cygwin for a couple of years and have only just gotten around to sorting out obvious niggles with Cygwin, don't hold your breath. I don't even use a Mac.

Had another query about the details of a rosette constellation; another subtle fix to a simulation got done. Getting phase offsets right between satellites' orbital planes can be quite tricky, and it's taken me a while to grasp how the constellations map onto the simulator. When the constellation has a lot of visible satellites like Galileo, it's hard to figure out from inspection what's right.

I'd be playing snowboarding games instead of typing this, but the family is watching television... the way to be productive in life seems to be to spend all the time you'd spend watching television staring at a computer screen instead.

A question from someone in China -- why is this a minus when this other script uses a plus? -- prompted me to look at my SaVi simulations, and made me realise that their coverage properties could be improved.

So I wound up checking all my Ballard rosette simulations, and correcting the Celestri and GS-2 scripts, because with the phasing reversed coverage was clearly much improved in them. (Celestri lost a couple of small coverage gaps that had always bugged me and now appears more regular, while GS-2 increases in evened-out diversity from mostly-one-satellite-overhead to mostly-two-satellites-overhead.)

A look at Skybridge and @contact couldn't come up with any improvement, so those rosette scripts are unchanged. But with many eyes all bugs are shallow, eh?

I'm wondering if I'll do as much work on SaVi development this Christmas as I have done in Christmases past. It's hard to get excited about anything on the SaVi bugs list.

43 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!