Advogato's Number On the Road: Paris

Posted 16 Feb 2000 at 09:23 UTC by advogato Share This

This week's installment of Advogato's Number is brought to you from United flight 999, Boston to SFO, written on my old laptop, the trusty spiral-bound notebook and Uni-ball Vision pen. My TP600 is grounded with battery problems and besides, they'll even let me run it during takeoffs and landings. I've been on the road for a couple of weeks now, and will be home for a month or so before Guadec.

The first week of this month, Advogato was at Linux Expo Paris. He spent the day before the conference strolling the city, observing the culture. Wine, cheese, and bread are very important to French culture (and Advogato certainly appreciates fine French cheeses!), but it seems to me that the culture is defined primarily by the French language. Having recently read Words and Rules by Steven Pinker, I kept an ear out for language clues. It is very clear, when you look, that French and English were once the same language - so many of the words are the same, even if they have diverged in shades of meaning.

Yet, they certainly are different languages now, and the Acadamie Française is charged with a special mission to maintain the integrity of the language, especially against the gale-force winds of the English language. Consequently, while most languages in the world have adopted English tems for computers and other technical fields, in French there is a new, parallel set.

For example, "free software operating system" is rendered in French as "système d'exploitation libre." "Système" is the same word as in English, and "exploitation" is similar, but without the negative connotation - a system for exploiting, or operating, a computer ("ordinateur"). Finally, "libre" means free, but in the sense of free speech ("liberty") rather than free beer ("gratis").

"Système d'exploitation libre" was very much the theme of the conference, in both the French and English senses of the words. Linux has made it big in Europe, as well as in the US. The show floor was packed with slick booths of large, powerful companies, eager to show off their impressive marketing budgets. Yet, most of what they were selling was plain, old-fashioned proprietary software to run on top of Linux. Advogato spent an hour or two wandering the booths, then was frankly bored - not much of the stuff being sold, or money being spent, is really advancing the cause of making a complete free software system into a richer, more powerful tool.

Nowhere was this contrast more striking than in Michael Cowpland's keynote address, which was basically an extended advertisement for Corel Linux and Corel Corporation. Cowpland spent a good part of the talk extolling the benefits of Linux and free software in general, including the robustness that comes from having so many eyes on the source. He then proceeded smoothly to promote Corel's proprietary software products running on Linux, such as Corel Office and Corel Draw. Advogato could feel the gears shifting in his head. For some strange reason, Cowpland did not feel it necessary to explain that, since these products are not free software, users shouldn't expect any of the benefits, such as the careful attention to reliability and compatibility so characteristic of free software.

Advogato has no problem with the existence of proprietary software, and it certainly makes a lot of sense to choose a solid, well supported operating system on which to run this software. But I do have a problem with the blurring between the two in the public's mind, and it seems to me that this blurring is being actively promoted by many Linux vendors. Linux is well known as a spectacularly successful example of free software, yet when people plunk down a wad of cash to buy some Linux, most of the time they're getting a significant dose of proprietary software as well.

Why does this matter? The conference session on clustering provided an excellent example. After the Mosix talk, an intriguing research project, Alinka and Suse presented their clustering solutions, both based on proprietary software. These systems probably work very well, and the people who buy them should no doubt consider them money well spent. However, I think it's fair to say that the SRO audience in the room was blown away by Peter Braam's talk on the work of the Linux-HA project, which is entirely based on free software. It's clear that this project intends to solvee the hard problems of making clusters Just Work, such as making making sure that systems go down and recover in the right sequence. They've come up with some very impressive stuff, including the Intermezzo file system and a general, scalable distributed lock manager. In the true spirit of academic inquiry, much of the work is based on previously publishe results, particularly the DEC VAXcluster systems of the early '80s. I don't see how this kind of work would be possible without the open cooperation of lots of bright people that free software brings. It seems likely to me that this project will bear sweet fruit.

The opportunity to build software systems that suck less is one of the main reasons I'm involved with free software. Yes, as Richard Stallman's talk strongly emphasized, the freedom is important too. Much of RMS's talk was taken up by a history of the Gnu project and a clear, compelling discription of the freedoms guaranteed by free software.

He then presented a plea that people call the complete operating system with the Linux kernel "GNU/Linux" rather than simply "Linux." Advogato has always been dubious of this request, because a Linux system contains many components from many different sources, each of which is essential for the system as a whole. In fact, the fact that the Gnu project adopted X as the windowing system is entirely analogous to RMS's assertion that the Linux project adopted Gnu tools. But nobody is insisting that Gnu be called "X/Gnu", for example.

It's easy to pick on RMS for coming across as a glory hound (or for having sour grapes about the relative failure of Hurd), but to do so misses an important point: calling the system "GNU/Linux" emphasizes the free software nature of the project, and brings the philosophy and goals of the Gnu project to the public's attention. In a time where many users are probably quite confused about the roles of free and proprietary software, this can only be a good thing. Maybe it's worth swallowing a bit of pride and using the term, if only for the pragmatic purposes of helping to educate the masses.

RMS described the requirements for being a Saint in the church of Gnu (alongside St. Ignucious himself); you simply have to eschew the use of proprietary software. Many in the audience were uncomfortable with this, at least until Mozilla ships.

My talk on libart went well, followed by an excellent presentation by Miguel de Icaza on new technologies in Gnome. He introduced his talk with a slide about Helix Code, the company he recently founded with his friend Nat Friedman, and clarified that "no, we're not lovers, and no, we're not gay." Ok, sure!

Overall, I had a great time in Paris, and was made to feel very welcome by French free software hackers, and I especially enjoyed meeting David Turner of FreeType fame and Mathieu Lacage, the organizer of Guadec (even if Mathieu did stick me with the check at dinner the first night!).

On the flight back, I was reminded, though, of the fact that our shared interest in free software doesn't completely break down the language barrier. After all, all the hackers I talked to spoke English. Even though Linux is a worldwide pheonomenon, it's still worth remembering that the free software community is dominated by the English language. One more reason why good, free documentation is important: it makes life easier for the hard-working translators, and makes free software more accessible to people who don't speak English.

I'm really looking forward to going back to Paris next month for Guadec. Hopefully, I'll see you there!


Yes, he paid :), posted 16 Feb 2000 at 09:59 UTC by mathieu » (Master)

Ok, I have no credit-card, I had forgotten my checks and I had no money on me.
I felt baaad. Hopefully, raph had a credit-card and he saved me from doing the washing.
So, thanks raph :)

language barrier, posted 16 Feb 2000 at 12:18 UTC by caolan » (Master)

Of course the English dominiation issue isn't just a free software problem, but a pretty widespread problem throughout the scientific and software industries. From a coding perspective one big suggestion I would have, now that I am coming at it from the other direction. i.e. reading german code, is that when you comment use full words.

It is relatively easy to look up a full words in a dictionary, but it can be really difficult to figure out an abbreviation.

C.

Internationalization focus a bit wrong, posted 16 Feb 2000 at 17:38 UTC by Radagast » (Journeyer)

Translations of the software in general is, in my opinion, not that great a problem. As English is taught to a larger and larger percentage of the world's population in school, it will become less of a problem. It's a lot more interesting to see things like Pango, which will let you write and display text in all languages. This is a true point for internationalization. Having the pulldown menus in every obscure language you can think of is not.

As for comments in code, and code in general (that is, identifier names), that should be in English no matter what, period. I'm not a native English speaker, but I can read and understand English quite well, and I'll be pissed off if code I have to work with is in Spanish (which happens quite a bit here, as well as MixedCaseIdentifiers, but I digress). If there is any chance your code is going to be looked at by anyone else at all, your comments and identifiers should be in English (and you should use accepted terms for things, which assumes a certain CompSci background, but that's a different issue).

exploitation versus operating, posted 16 Feb 2000 at 19:29 UTC by mathieu » (Master)

raph, I would like to know what you imply here. quoting you: ""exploitation" is similar, but without the negative connotation - a system for exploiting, or operating, a computer".
What negative connotation is involved in "operating" ? I probably have missed something here...

A little followup, posted 16 Feb 2000 at 20:12 UTC by raph » (Master)

I was talking about the negative connotation in "exploiting." "Operating" really has no negative connotation in English.

I've definitely experienced the language barrier in the other direction. For example, the GYVE project obviously has some very talented people working on it, and was started well in advance of Gill. However, I was very frustrated in trying to communicate the differences in vision, particularly that I thought the libart imaging model had a much brighter future than Display PostScript. So we ended up going our separate ways, and another possibility for collaboration was missed. I have a feeling things could have gone better without the language barrier.

Yet, I also feel that forcing the free software community to use English is arrogant and imperialistic in a way. Monocultures are not good in ecology, and I can't imagine they're any healthier in free software. Yes, science has the same problem, but science is basically a sport for rich countries, or at best a small elite within poor countries. I expect a lot of people around the world to be using Linux and other free software over the next few years.

Forced to use English?, posted 16 Feb 2000 at 21:21 UTC by prozac » (Journeyer)

Yet, I also feel that forcing the free software community to use English is arrogant and imperialistic in a way.

Do you really see it that way? The forcing part I mean.

As with science and technology so with computers. I once worked for a Danish company (Danfoss), and when asked why all of their senior staff were required to speak English, and why much of their documentation was is in English, the response was, "We have to." But not in that they were being forced to, but in that, since Danfoss is a world wide business, all of their offices have to all communicate together, communicate very specific and unambiguous things, such as schematics and notation. It is much more efficient for all to communicate in the same language rather than having each office learn the laguage of every office.

I was involved in a meeting in Sonderburg with representatives from France, Germany, Japan, Korea, Sweden and the UK (I am from the US), and of course Denmark. We had to all speak the same language or things would have been very difficult.

Of course it makes it easier for people of English speaking countries. And I really feel, well, left out by not speaking other languages. But I do not feel lazy as many have labeled Americans because English is the most common language in so many technical things. I think it just worked out that way.

nitpicking poor arguments, posted 16 Feb 2000 at 22:45 UTC by rillian » (Master)

RMS described the requirements for being a Saint...you simply have to eschew the use of proprietary software.

This is getting off topic, but one of my biggest complaints with RMS is his refusal to apply the ideas of free software to anything but his particular definition of software. It's nice to say that you shouldn't use proprietary software; I agree with him. But I'd also like the hardware my system runs on to be a product of free software. When I ask him about this possibility he laughs it off, saying "there's no such thing as a car copier." True enough, but not an argument. I think he's wrong, and that bothers me much more than any possible 'glory hounding'.

I'm not convinced that the lack of desktop chip fads is all that different from the situation 15 years ago, when RMS quit his job at MIT to write free software--on a proprietary workstation owned by the university. Of course the issues are different, and the only way I seem "everybody" buying a chip fabricator like they're buying computers is if it also happens to make coffee, decorative craftwork, and everything else. But that doesn't mean an interested developer couldn't build one in the garage, based on cheap prototyping techniques developed in the free by a community, and amortize material and equipment costs by running things for others locally. We've done that model in software already. It's called a mainframe. Remember "5 or so computers should be enough for the nation's needs"?

API are designed with the english langage in mind., posted 18 Feb 2000 at 12:26 UTC by mathieu » (Master)

One interesting thing to note about code is that APIs are usually designed with the english langage in mind (actually, I NEVER used any non-english API). As a consequence, code which uses non-english variable names and function names and english APIs should be banished.

I have been obliged to use some code from some networking profs here, at uni -people who should NEVER be asked to code anything but that is offtopic-, and what was the most difficult was to get used to the intermingling english API calls with french variable names (although I am fluent in english and french is my native langage). It is extremly disturbing and requires a constant mind-change to read such code -actually, the code was even worse than just this but... So many buffer overflows...-

Although: now, I recall using a graphics library in french. It was one of my worst experience but I think this was mainly due to the fact that the API was simply badly designed.

In order to be a good programmer, one must learn English (alas?), posted 18 Feb 2000 at 17:18 UTC by Raphael » (Master)

First, a bit of background about myself: sometimes, I like to describe myself as a French-speaking Belgian guy working in English in Germany for a Swedish company (and driving through the Netherlands in order to go from my home in Belgium to work in Germany). I probably speak and write in English more often than in French, mostly because I work in a big company in which a good knowledge of English is required and also because I still do not know enough German so I use English when I talk with my German co-workers.

As a researcher and developer working for an international company, there is no question about what language should be used when writing code or documentation: everything must be in English. It is likely that the next person who will maintain the code that I am writing now does not understand my native language and I probably do not understand hers, so English is the only way to ensure that we can communicate.

But even if my job would not require it, it would not make sense for me to write my code in a language other than English. Even if I write some code for myself and I do not intend to give it to anyone else. Most of the APIs that I ever had to use are using English names for all functions, types and classes, so it looks very bad if I mix that with French names. It could also be confusing, if you consider that the French translation for "a queue" is "une file" and the translation for "a file" is "un fichier". You can get in trouble if you read some code that contains function calls like: queue_append (file, fichier); or if you are not sure about the best mapping between the words in the two languages: draw_outline (calque, fond, bord);. As others have mentioned, it gets even worse if you start using abbreviations.

But on the other hand, some non-English CS students are faced with the opposite problem: they do not know English well enough, so they try to use their native language as much as possible in the code in order to be able to understand it later. A good friend of mine once told me something like: "It is already hard enough to understand the API written in this cryptic English language, so why should I make my task even more difficult by naming my variables and functions in a language that I am not familiar with?" Well, he was right to some extent. Besides, as a student, you often have to write your code and / or your comments in French anyway (or in German, Spanish, or whatever is your native language) because the other students and even your professor could think that you are a pedantic defender of the English language or that you simply copied the code from a book or from a more experienced programmer and you forgot to rename the variables appropriately. Don't laugh, I know that it happens...

Looking back at the first programs I wrote (good old ZX81, then ZX Spectrum), I remember that I started with something like: "code and comments in French," then later "code in English and comments in French" and finally everything in English. One must be reasonably familiar with the language before being able to name the functions and variables correctly (and also being able to re-read one's own code later). Learning a language takes time, and this applies to programming languages as well as natural languages. It would be foolish to require all CS students to use English exclusively in their code: many would fail or would be disgusted.

But once you are past the first few learning steps, it is vital to use English for everything. Because at some point in time, you learn more by reading, modifying and trying to understand code written by others than by reading books or writing programs in isolation. As soon as you have to share code with others, English is the only good choice. Besides, all the good reference books are written in English. Even the translated versions contain enough references to English terms that you could not read them if you do not know English.

This is especially true for Open Source programs: reading and writing OSS is an excellent learning tool. This is a good opportunity to share code, tips and tricks with experienced programmers from all around the world. But it is only possible if everything is written in English. Releasing a nice Open Source program written entirely in French is probably not more useful to a Japanese developer than if the source code had not been released at all. And vice-versa.

I believe that beyond a certain point, you cannot gain more experience if you do not read and write code in English. And since Open Source is an excellent learning tool, I would conclude with: In order to be a good Open Source programmer, one must be good at English.

Hmmm... That was a long comment for a short conclusion. I will try to write shorter comments next time... :-)

.. and it's not just programming, posted 28 Feb 2000 at 00:03 UTC by jennv » (Journeyer)

Programming is not the only field which uses a specific language. Ballet uses French for its technical terms. Music uses Italian.

Admittedly, these don't require fluency in a different human-language just to work in the field. I felt it appropriate, however, to point out that there is precedent - and precedent other than medieval scholarship!

Jenn V.

Internationalization of software is indeed necessary, posted 29 Feb 2000 at 16:51 UTC by monniaux » (Journeyer)

I disagree with an earlier comment that pretended that translating of programs (menus etc...) was unnecessary, since today young people in most countries are learning English.

I think this comment disregards the fact that, even though there have been improvements in foreign language teaching, people, even young, even in highly developped countries, do not speak English well. They are more often than not able to complete simple tasks such as asking for directions, but are not able to read English documentation.

Do I exaggerate ? No. I teach Java and basic algorithmics in a French university. My students asked me for documentation. I pointed them to the official site for Java. They just could not believe no documentation was available in French. Sure, the problem might be that English is taught as a foreign language, not as a language you need for day-to-day activities such as programming. A segmented view of education is surely responsible for this "English belongs in the English class" thought.

Nevertheless, that is what happens in real life at some allegedly good university in a major city in one of the richest countries in the world. Now imagine the situation in the middle of countries that cannot afford the luxury of running extensive education systems.

We in the technical community have become so used to writing programs, documentation, e-mail and WWW pages that we do not know anymore that for normal people out there, English is just a faint souvenir from highschool days It is the case for my mother, for example.

Providers of commercial proprietary software have long understood the need for localized software. Let us not fool ourselves.

The problem with the internationalization and localization of software is that it is often inadequate and amateurish. Often, the spelling of the translated strings is approximate ; some translations in Gnome or the GNU Libc show that the translator did not have proper command of English and/or the target language.

Is there any forum on internalization issues ?

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!

X
Share this page