Hacking for Christ

Decent Editor Wanted

For a long time I've used NEdit as my editor. It has a number of annoying quirks, but it's just about usable. Having just upgraded my desktop to Ubuntu, and found that NEdit compiled with LessTif has even more bugs, and looks really ugly (and I can't work out in which file to put the X resources given in the beautification tutorial) I'm thinking about a replacement.

Given the general usefulness of editors, and the fact that everyone and his mother seems to have written one, you would have thought it would be quite easy to find one which meets my simple requirements. They are:

• Currently maintained (i.e. not abandonware)
• X application, decent-looking
• Good syntax highlighting
• Supports, or can be made to support, the key bindings embedded in my brain like Ctrl-C for Copy, Ctrl-V for paste, Tab to indent selection.
• No vi-like modes. Modal interfaces are bad.
• Supports standard Perl regexps in the search/replace box

Here's what I've tried, and why I've rejected them:

• nedit:
  • Looks like a dog in Ubuntu
  • Copy and paste to other applications doesn't work properly
• gedit:
  • No regular expressions
  • No key rebinding
• SciTE:
  • Doesn't support proper regexps
  • Spews pango errors into the console I launched it from (really irritating)
  • Doesn't notice when files have changed on disk

XEmacs is, I suppose, a possibility, but I'd need to spend an age converting its 1970s keybindings into what the rest of the world uses today. There are some good editors for KDE, but I'd rather not load all the KDE libs on my GNOME desktop just to get an editor. I suspect Eclipse would probably be far too heavyweight.

Anyone got any other ideas, or know anything I've missed about some of the above? If you comment, please let me know what editor you are using even if it doesn't fit my requirements.

Tiny Bug, Big Tiny Effect

I just fixed a teeny tiny bug in my relicensing script.

While recursing, it was checking directories against the excluded list by doing an endsWith(). However, one of the directories to be excluded was "ef" (Electrical Fire), an obsolete piece of debugging code, and it turns out that this was matching all directories ending in "ef" - like, for example "pref".

This small fix has revealed 3500 more files to be relicensed that heretofore I had not suspected the existence of. In fact, I only noticed because I found a freaky file with an NPL header but MPL text, and did an LXR search for "NPL 1.1" to see how many like that there were. Anyway, hopefully my script will make fairly short work of them.

Today's lesson then, children, is: Don't trust new tools absolutely. Run cheap cross-checks.

[Update 5.47pm]: OK, so the fix had a bug. The actual number of files is 91, not 3500. Second lesson for today: don't be too eager to write blog posts explaining your own foolishness, lest you demonstrate it to the world in unintended ways...

Web 1.0 vs Web 2.0

mpt doesn't blog much, but it's normally worth waiting for...

Microsoft Fine Could Be A Price Worth Paying

My latest article for The Times Online is now available.Called "Microsoft fine could be a price worth paying", it talks about recent developments in EU vs. Microsoft.

I'm a little less sure of all the precise technical detail in this one, so please add a comment if I've got something wrong.

Chemistry Of A "Therm au Rouge"

This may be the wrong sort of blog on which to ask this, but...

Our family has a "Therm au Rouge" (that site has the best images) wine-warming sleeve - one of those things you buy for a Christmas present out of desperation. It wraps round a bottle of wine and, after "clicking" a small square of metal inside the pouch by bending it, an exothermic reation is initiated which warms the wine. The reaction spreads out from the "clicker" and the clear liquid in the sleeve turns into a white solid, emitting heat in the process.

The reaction is reversible; when it's complete, you can boil the pouch in water for ten minutes to reverse it and turn the liquid back to clear. The instructions say it'll work at least 100 times before you should replace it.

I've searched the web but I can't find out:

1. What is the chemical reaction involved?
2. How does the clicker which initiates it work?

Our family is having a Christmas "discussion" about this so if anyone has any ideas, I'd love to hear them...

Put Yourself On Frappr

I just put out a call in netscape.public.mozilla.l10n for localisers to put themselves on the Mozilla map on Frappr (see below). But everyone all Mozilla project code or QA contributors should feel free to add themselves - the more the merrier!

Dear mozilla.org localisers,

We'd like to make a map of where you all are. This is because the Foundation is considering putting on some events or meetings for localisers, and this information would greatly help with the planning.

So could everyone who is in a localisation team take a moment to do the following? If l10n team leaders could pass this message on to their members, that would be fantastic. The more info we have, the better we can plan. We are using Frappr, a new Google Maps mashup which lets people make community maps.

1) Sign up for Frappr at http://www.frappr.com/ - all you need to supply is a name, email address and password, and you can use the site immediately.

2) During signup, in the "Occupation" section of the Basic Profile, put the following text:

l10n products: <product>; <product>
l10n language: <code> (<free text description>)

Example:

l10n products: Firefox; Camino
l10n language: en-GB (U.K. English)

The products should be taken from the following set:
"Firefox, Thunderbird, Seamonkey, Camino, Bugzilla"
and semicolon-separated. If you do multiple languages, again please use a semicolon as the separator.

You can say "product" or "products" or "language" or "languages" depending on which is correct; I think the regular expressions in any future automatic parsing of this information can probably cope with that. ;-)

3) Go to http://www.frappr.com/mozilla

4) Scroll down the page to "Add Yourself" and click "Add Me!"

That's it! Thanks!

Gerv

[Update 2005-12-23 5.21pm GMT] I've made it clear that the map is intended for code/QA contributors. If e.g. SpreadFirefox people want to make one, that's cool, but please don't add yourselves to this one as there's tons of you and it would make it really hard to see the developers!

How Sad...

I just signed up for an account on Frappr, which is a service which allows you to create a map of the geolocation of your friends or other communities. There's a Mozilla community; do sign up and join in.

Having added myself to that, I clicked on "Gervase Markham's Frappr" at the top of the page. I was presented with a page that had a map of Europe on one side, bare apart from a solitary blue pin stuck in London, and a big heading on the right side which said: "Currently you do not have any friends."

LOL! They could at least have said "...friends registered on Frappr"!

Calling Q Companies...

"Q" is the only letter for which we do not have a company using Bugzilla among the 478 listed. Any companies out there beginning with the letter Q who are using Bugzilla want to be added? :-)

Inside Gerv

As some of you may remember, I had a couple of operations earlier this year to remove what turned out to be the primary site of my cancer from the floor of my mouth. (For those just catching up, I have Adenoid Cystic Carcinoma (ACC), which in my case originated in a salivary gland.)

However, the scans done around the same time revealed that I had a number of lung metastases. A metastasis is a secondary cancer caused when the primary sheds cells which implant themselves in other parts of the body; this is a characteristic of most malignant cancers. Further scans done in October confirmed this; I have five in the left lung and two in the right so far visible. There may well be more which are too small to show up on the scan.

I'd been hoping to share these images with you before now, but the files on the CD-ROM I have are in some strange medical interchange format that even imagemagick doesn't understand (although it claims to). So I had to fire up the proprietary viewer on a Windows machine and take a screenshot.

This is a cross-sectional CT picture of my lungs, taken at the beginning of October this year (click the image for a larger version). The lungs are the black areas - they normally don't contain much that shows up on a CT scan. This cross-section is from near the top; the bony areas (white) are the top of my ribcage, my spine and my shoulder blades. My left lung is on the right of the picture, and the lump in the bottom left corner is the largest and most clearly defined of the seven metastases. It's about 1cm across.

A combined PET-CT scan showed no discernible metabolic activity, so they aren't growing very fast. However, one is getting close to a major blood vessel, so reasonably swift removal is recommended. So, I am having two back-to-back operations - one for each lung - probably just after Easter 2006. Each one will take about a month, including recovery time.

How am I feeling? Well, I was in the pub on Wednesday night with a bunch of Ruby and Perl geeks. One chap (whose name I forget, and who may not want it publicised anyway!) was talking about how he thought for a while he had testicular cancer, and paid £200 to go for a private scan, rather than wait two weeks to get it free on the National Health Service, because he was so stressed about the possibility. In the end, it turned out to be a benign cyst. I then mentioned that I had cancer with lung metastases, and was probably having two operations this year. He said "But they can't be malignant, right? Otherwise you would be worried. And you don't look worried." I said "The creator of the universe is looking after me personally. What possible reason do I have to worry?"

Which Scam Is It?

I got an email which smelt like a circular, and began as follows:

On behalf of the WMSCI 2006 Organizing Committee, I would like to invite you to participate in the 10th World Multi-Conference on Systemics, Cybernetics and Informatics (http://www.iiisci.org/wmsci2006) , which will take place in Orlando, Florida, USA, from July 16-19, 2006.

The best 10%-20% of the papers will be published in Volume 4 of JSCI Journal (http://www.iiisci.org/Journal/SCI/Home.asp). 12 issues of the volumes 1 and 2 of the Journal have been sent to about 200 university and research libraries, and 6 issues of Volume 3 (2005) will be sent to a larger number of library. Promotional, free subscriptions, for 2 years, are being considered for the organizations of the Journal's authors.

We are emphasizing the area of Information Systems Development which is related to your specific area...

The question is less "is this a scam?", and more "which scam is it?" Is the idea to get money out of you for presenting a paper to an empty room? Or is it so people can sign up for it and use the bogus kudos ("I presented my work at the WMSCI 2006 conference, and was published in the JSCI Journal") to try and gain credibility? Or does it exist so that people can tell their bosses they are going to a conference, then take a week's trip to Florida? Or all of the above?

Times Article List

I've put a new page on gerv.net listing all the Times articles I've written so far, in case anyone is enjoying reading them but missed one.

The Battle for Bangalore

The latest "Our Man from Mozilla" column in The Times Online is now available. Titled "The Battle for Bangalore", it contrasts two recent events in the city in terms of the battle for the hearts and minds of legions of Indian software developers.

Old DVD Drives Wanted

The Windows Vista team are going to stop supporting "old" DVD drives. By "old", they mean the ones which have no hardware region checking and so can be made multi-region simply by applying software hacks. The main reason for this, apparently, is that they don't have very many working drives like that any more.

Clearly, this has nothing to do with the entertainment industry trying to get more of a lock on world trade in DVDs. So, I call on everyone who's got a pre-2000 machine sitting around with a DVD drive they don't use to remove it, package it up and send it to The Windows Vista DVD Hardware Support Testing Department, Microsoft Corporation, One Microsoft Way, Redmond, WA, 98052-6399, USA. I'm sure that, if we all did that, they'd keep supporting our hardware. Right? :-)

Worldpay Recommends Firefox

Worldpay are an online payments solution for small businesses, and they recommend Firefox. You can't see the recommendation by default but, if you go to their page and turn off page styles (View | Page Style | No Style), as if you were using a browser that doesn't support CSS well, such as Netscape 4, you see:

Why does WorldPay look like this?
Your browser cannot find our style and presentation information. You are welcome to use the page as is, or upgrade your browser to its latest version. We recommend using Mozilla Firefox which can be downloaded from here. If you are using another browser, please see your browser's website for more information.

Cool! I'm glad some financial institutions are cottoning on to the fact that more Firefox use is to be encouraged.

Resource Exhaustion Tools

I'm trying to test a normal Windows desktop GUI app, and want to see how it performs in CPU-starved/low memory/low disk space conditions. I've been looking around for test tools which will eat all the CPU or memory at the press of a button, or fill your disk up in a nicely reversible way, but I can't find any. It would be nice if it also allowed you to easily simulate a network disconnect, either from all machines or a defined subset. All the "QA software" I can find seems to be entire testing frameworks, or is for web app load testing.

Does anyone know of utilities which do roughly what I want? I would hope that many QA departments test their software in such conditions, and so other people will have encountered this problem before me...

Open Rights Group Won't Take My Money!

Like 1033 other people, I signed up for the PledgeBank pledge:

I will create a standing order of 5 pounds per month to support an organisation that will campaign for digital rights in the UK but only if 1,000 other people will too.

(Actually, I was so keen I signed it twice without noticing, so I guess I have to pledge £10 per month...) The pledge has been successfully completed, and the organisation is the Open Rights Group (ORG). However, nowhere on their site does it tell me where to send the money! I've emailed Danny O'Brien to ask him what to do, but have heard nothing back.

Please, ORG, take my money! I'm begging you! :-)

UI Problem: Dynamic Locale Switching

I recently came across bsmedberg's Locale Switcher extension. I was thinking that something like this was needed for Internet cafes, where successive people may want to use the browser in different languages. However, the ability to change languages on the fly is not common, and so the UI for doing so needs to be really obvious and up-front if people are going to find it and take advantage of it.

The problem is further complicated by a technical limitation. It's not possible to change the language of an already-existing window. All you can say is that all new windows should have the new language. And so the UI can't just be a set of buttons where you click one and get on with what you were doing.

Here's my current attempt at the UI for an Internet Cafe Locale Switcher. Comments are very welcome.

Each browser window has a toolbar, which looks something like this (except fixed up and fitting with the rest of the UI and so on):

• The toolbar appears in every top-level window, and changes to it are synced between windows.

• As you hover over each control in the toolbar, two things happen:
  
  
  
  1. The phrases "New windows in" and "Open New Window" change to be in the relevant language
     
  2. A tooltip appears, saying "Make new windows use an English interface" or equivalent.
     

• The space for the "New window in:" text is long enough for the longest configured string, so that the controls don't move during mouseover.
  

This is the clearest thing I can come up with which reflects the fact that the change of language is really a setting for future windows rather than for the current window, yet makes it as easy as possible to get a new window in that language, without possibly losing or hiding any work the user may have in the current window.

Comments?

Scripts and Barriers

I'm currently sitting in the ICANN IDN workshop in Vancouver, Canada. It's past 7pm now, and we've been here since half past twelve. It's been very interesting, if long. The main focus of the session is IDN at the top level - IDN TLDs - but we also had an hour on IDN at the second level, where Michel Suignard of Microsoft and I gave presentations on IDN in IE and Mozilla respectively.

One feature IE's IDN implementation has which we are not proposing to adopt is that it has a notion of "scripts that you know about", and if you visit a website using an IDN in a script you don't know, it displays the IDN as punycode and gives you a yellow security warning bar.

I think this idea is very worrying, because it balkanises the Internet. Instead of a global namespace, you have lots of smaller ones with mutual distrust between them. This seems to me to be counter to the very idea of IDNs - that everyone in the world is placed on the same level, rather than the current situation where Latin-based languages and English are the first-class citizens. It would also have a deterrent effect on IDN uptake. Who is going to use an IDN name if an unknown percentage of web surfers are going to get given a security warning when they visit it? This is particularly bad for users of minority scripts, who almost no-one is going to trust be default.

From an implementation perspective, it also has practical problems. How is an Iranian going to react when he visits an Internet cafe in the US and is told that all his websites are suspicious? What about computers used by multiple people, each of which knows a different set of scripts? (E.g. a house where the father happens to have learnt Chinese at night school, but the rest of the family don't know it.) It also means that your browser needs additional configuration because of IDN, which I believe should be an active anti-goal of our IDN work.

But I also can't see the point. What risk does it mitigate? As long as it doesn't look like the domain name of my bank (and there are other mechanisms for dealing with that), where is the risk for me in visiting a domain whose name happens to be in Russian, or Chinese? I can't see one.

I hope that Microsoft change their mind and remove this feature.