Older blog entries for bolsh (starting at number 73)

On gimp-users today someone asked a question about the availability of panorama tools for the GIMP.

I had previously searched for and found such a plug-in, but it had been a couple of years, so I went on a quick web search to see if I could find the one I remembered.

I uncovered this massive community of people interested in stitching tools, with a very active mailing list and several applications (some GIMP plug-ins too) - I must admit I was surprised, since I had never heard of most of these applications before.

Here are the results of that "quick" search, for posterity.

  • The original set of tools I remembered was pandora. It has its own webpage with a more recent version to download: http://www.shallowsky.com/software/pandora/

  • Or it might have been Panotools. The home page for Panorama tools is http://panotools.sourceforge.net/ and it is GIMP 2.0 compatible.

  • However I found a whole mailing list which is *very* active on panotools frontends - http://www.email-lists.org/mailman/listinfo/ptx

  • There are many programs mentioned on that list, among them hugin seems to be the most reccommended: http://hugin.sourceforge.net/

  • And the enblend program: http://enblend.sourceforge.net/

  • And finally autostitch: http://user.cs.tu-berlin.de/~nowozin/autopano-sift/

The ptx mailing list has about 100 mails a month, primarily from hugin developers. This is a great model for "themed" software - different packages whose functionality complements each other.

Wow. It's amazing what years of fear will do to a country.

We need more rebel movies. Bring back Dean and Brando!

In other news, Jean Pierre Raffarin is trying a new tack to getting fired. I'm not against allowing people to have longer working weeks, but I'm in the minority.

That's because the way things currently work in France is that you have the same working week as everywhere else (40 to 50 hours), and you get 10 to 20 extra holiday days per year. Last year I had 40 days holidays, this year (because I changed jobs) that'll be down to about 15 (10 days RTT and 5 days holidays).

The French have this weird system where you have to earn your holidays in one year, and take them the year after. Then when you quit you are paid for all of the holidays you earned, but didn't have the right to take yet.

Wiki news

On a thread on marketing-list last week, a point was made that right now there are too many avenues of communication into the GNOME project. Many requests for help appear to fall on deaf ears, simply because they are never seen by the people best able to respond. With the vast number of mailing lists, websites and other resources, it can be hard to know where to turn for help.

That was part of the reason why gnomesupport.org was so successful as a help site - people could turn up to the forum, ask questions, and expect answers. One of the things which gnomesupport.org did was put up a wiki, long before live.gnome.org existed.

So as a small step towards the consolidation of points-of-entry for low-level users, I proposed to stro over at gnomesupport that we work towards migrating all of the information from the gnomesupport pages to live.gnome.org. We haven't worked out all of the details yet, but at least there is agreement that this is a good thing to do. The GNUCash developers, who didn't even know about live.gnome.org, used the gnomesupport wiki exclusively, and a first response suggests they are happy to move too.

So, it's a small step, but soon there will be only one wiki for GNOME, which is a Good Thing. Many thanks to Claus Schwarm for pointing out the gnomesupport site to me, and to stro for helping work out the details of the move.

Robert: Your argument against rent control is a good one, but it has a fatal flaw. When you say

the problem is that, in a stable market, prices properly reflect the position of supply versus demand

you're correct. However, accomodation is not a completely open market, and thus purely market forces do not suffice to stabilise it. Accommodation is a limited resource, and usually people have little choice where they live once they take a given job.

The situation in Dublin in the '90s is a good example. At the start of the decades, rents were cheap by European standards. Even in 1996, when I moved to Dublin, I was paying £35 ($50) a week for a bedsit, which became £40 ($57) a week in a house-share soon after.

Then the job market exploded. The population of Dublin and its greater metropolitan area went up from under a million in 1991 to 1.4 million (including satellite towns) in 2002. My rent went from £35 a week to £450 ($600) a month when I moved to a 40 m2 (400 sq ft) apartment in 99, and from there to £750 (over $1000) a month when we moved to a slightly bigger apartment - 55m2 (550 sq ft) in 2000. That's an inflation rate of 300% or 400%, which was caused by a mix of market factors and geographical factors.

  • More meople were arriving in Dublin than there were places to accommodate them
  • Dublin's design (and planning laws) meant that it was hard to accommodate everyone in the town, leading to some urban sprawl
  • Houses were coming on the market 6 to 9 months after they were needed, leading to a tulip war over new houses
  • People needed accommodation, so to get what little there was, took bigger loans at the bank
  • The banks, seeing the market explode, were happy to give people more money than they could afford to pay back easily.
  • More and more people were buying property as an investment, raising rents to pay back mortgages
  • Most of the people migrating to Dublin were in the 20-30 age bracket, and most of them could not afford to buy at market prices

The classic laws of supply and demand make assumptions about buyers and sellers (such as, if a buyer is charging too much for a good, the seller can decide not to buy it) - the situation in the real world where a good is both a necessity and in limited supply mean that controlling the market is sometimes a good idea.

Just to clarify what I mean by rent control: when you agree to let a property at a specific rent, you are limited in the frequency of rent increases, and in their amount, by the law. Typical restrictions might be "no increase for 2 years, then no more than 3% per year". So if you rent an apartment in 2000 at $500 a month, then you know that your rent will not be any more than $515 a month in 2002.

In situations where rent control is in place (such as France), market pressures can still force rises in property prices and rental prices, while protecting existing tenants from landlords. Not that that's a bad thing, but it means you can't arbitrarily raise the rent to get in a "better" tenant, putting some poor 70 year old woman who worked all her life embroidering tablecloths out on the street.

A first: I needed information on the modbus or modbus+ protocols for work today (it's a low-level protocol common in embedded stuff - basically a master says to a slave "Give me 50 bytes starting at offset 0x08", and the slave does), and came across this page.

What's particular about the page is this line, in their product specs:

  • Straight forward per-developer-seat software license, no royalties or GPL

This is the first time that I have seen "no GPL" listed as a "feature" of a commercial product. I suspect that it's important in the context of an SDK,and if it's there it's because clients have asked the question, but it just struck me as odd. I guess it means that

  1. Free Software has mindshare even in niches where you wouldn't expect it, and
  2. the first reaction to Free Software is fear and uncertainty.
Luis: Of the 40 patches which got "reviewed" , over half were from the GIMP. Sven and mitch do a great job of looking at and commenting on every patch we get, but they don't always update the patch status. So all that had to be done was to go through the bugs, and mark the patches "needs-work", "committed" or "rejected" depending on the previous feedback. Easy peasy ;)

So can you update your list of "top performers" now? :)

Alan Horkan mentioned this presentation, which gave me something of a buzz. It's a presentation about and around social software, but is also valid for a bunch of other group situations.

I got particularly excited about the IRC/Wiki/conference call situation he described later on.

The author is discussing the design of social software, but he comes up with a number of insights that are somehow related to other things two:

  • A group is its own worst enemy
  • You can't separate social and technological issues
  • You need barriers to participation

Some related articles are here (Joel on Software) and here (LambdaMOO changes direction).

19 Jan 2005 (updated 19 Jan 2005 at 19:41 UTC) »
Wiki update

I added an "Upcoming events" section to the UserGroups page in the GNOME wiki and added Solutions Linux and Fosdem as two upcoming events where there will be a GNOME France presence.

I think that this kind of section is essential to make that page "live". User groups aren't just contact information, and many user groups have a presence at local conferences which most people don't know about. So tell everyone about it - you never know who might drop in.


Thanks to Damien Sandras, there is a GNOME developers room set aside in Fosdem which GNOME people are free to schedule as they want, but currently there are no talks planned for it - there are 3 KDE talks planned.

I'm not sure who the person to contact is to be able to add and solicit talks (perhaps Damien could say?), but Fosdem is a great conference, and it would be good to have 3 or 4 GNOME related talks in the developers room.

Update: Dodji Seketeli is looking after the GNOME developers room for Fosdem, and so far several talks are planned, but the page hasn't yet been updated.

So there seems to be some hullaballoo about Evolution being ported to Windows.

I agree that this is a Good Thing. The more choice we give people, the better off they, and we, will be. Plus, there is no doubt that porting Evo to windows will get it a larger user base, and some corporate acceptance. And it'll remove lots of artificial barriers against having GNU/Linux or Unix on the server side. So all in all, yay!

But... (what follows is an experiment with "Stream of Thought", so I may end up disagreeing with myself. This is healthy)

The main argument of people against this kind of thing is that if people can use all their free software applications on Windows, there's no incentive to change platforms. Proponents argue that when the only piece of software that you have to pay for is your OS, the choice is obvious and you switch. So when you get into TCO fights with Windows people, they can now say "thanks for all the groupware, you have just reduced our TCO, making Windows an even more attractive proposition for the enterprise". But then wa can reply "Sure - but we've also reduced the TCO for migrating to GNOME, since all those enterprise users don't need any retraining any more".

Of course, Microsoft will never argue that you can use free software on your Windows machine to reduce TCO - that would be hari-kiri of the highest order. But if they did, then I guess we have a counter-argument.

Also, in related news, Chandler, the Mitch Kapor backed groupware solution modelled on Domino, as far as I can tell, has had a change of direction. The guys over there have realised that big milestones which nobody uses have been slowing them down, so they are moving to what they call a "dogfood" model of development. That means that with their next version, 0.5, you will actually be able to do stuff like calendaring. Good work, guys! I hope that there has been some thought to coming up with an open standard for calendaring to share with the Evo and Sunbird projects (but I guess that goes without saying).

GUADEC update

Phew! Submissions "officially" closed last Wednesday, but we're nice guys, and there has been a trickle of late submissions coming in since then.

We have not been able to get back to everyone yet, but let me assure you that we are not going to refuse papers just because they were late. However, after the weekend, the selection committee are going to start separating the wheat from the chaff, and once that starts happening, papers are more likely to be rejected.

There are 56 submissions so far for the papers track, including tutorials and BOFs. It's not bad, but I thought there would be more. Perhaps it's a little early in the year for people to think about GUADEC.

We will be looking for less formal things (BOFs, flash talks, module maintainer brainstorming sessions and the such) a little closer to the conference (before the end of March) for people who really want to be on the schedule. And I expect that there will be a bunch of ad hoc impromptu sessions during the event too.

Thanks to everyone who submitted papers, sessions, BOFs and tutorials!

64 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!