Wed, 28 Jun 2006
Weekend in Znojmo
Last weekend I was with Iva and my coworkers on a trip to Znojmo. It was a nice trip and I am sure Iva liked it. I took my bicycle with the child seat, and we went for a ride to Šobes.
Iva seems to like riding with me in the child seat - I think she has much better view from the bike than from the car seat. I bought the seat with reclining back, so that she can sleep in the seat if she wants to. However, with a helmet, sleeping even in this type of seat is not very comfortable.
Thu, 22 Jun 2006
I have just came across the Perl6 Synopsis, so I wanted to see where the Perl development is heading to. I must say, I think moving to Perl6 would be a horrible task. I still did not read everything, but so far it seems some design decisions are a bit weird for me:
- Single dot (".") instead of an arrow ("->") as a dereference operator. While it is not bad per se (several languages like Ruby or C use it), I think the arrow is more readable (because it is wider, it provides better visual delimitation).
- Long dot (the "object\ .method()" construct). It is visually awful (as backslash is usually associated with string interpolation or escaping special characters). Whitespaces in Perl6 will be more involved as a part of syntax than in Perl5. I think the significant whitespace is a domain of Python, but now it is also in Perl6. I think introducing the long dot into the Perl6 syntax is probably connected with having to use dot as a dereference operator instead of an arrow.
- Mixed-case words (such as builtin type names, but I have seen this in other contexts in the Synopsis documents as well). IThinkThisIsAJavaWayOfDoingThings, WhichIsClearlyUnreadable. DefinitelyLessReadableThan underline_delimited_identifiers.
- Native versus object types (at least they have autoboxing). Having to distinguish between int and Int data types is not Do What I Mean. I think the main purpose of scripting languages is that the programmer does not need to think about (obvious) type conversions. Is it another try to sneak Java drawbacks^Wfeatures into Perl?
That said, there are definitely positive aspects of Perl6 as well - for example the ability to use any Unicode pair characters as comment or regex delimiters (for example the Japanese corner quotes in: "qr「match」"). But I think for me the above problems mean that I would have to stick with Perl5 as long as possible, or I would have to lear another language, which would have worse (or smaller) libraries than Perl has (Ruby), or which has even more annoying syntax (Python, the No-oneliners-please language).
Fri, 16 Jun 2006
People at manga.cz chat have told me about AnimeNFO Radio - a Shoutcast-based streaming radio-like application. It has a web-based front-end, in which users can request songs to be played, rate the songs, etc.
It mostly works, but there is a serious drawback - when I start to listen, and request some tracks to be played, they get added to the end of the queue (which is usually between half an hour and hour long). So by the time "my" songs are on air, I usually have to leave for lunch or something. I think it would be better to have some local player, which would know what tracks I have and which ones I prefer, and it should then create a mix of the local tracks with some remote streamed songs, preferably those recommended by usesr with similar preferences as I have. It could even send back info about the tracks I listen, and update my profile on the server.
Ther playlist lacks few tracks which I like: they don't have Nadesico OST, and they don't have the OP from Scrapped Princess, for example. I have yet to figure out how to posts the requests to their forum. On the other hand, I have already heared few interesting songs I don't have.
Tue, 13 Jun 2006
Cancel this account!
The dwmw2's story about his Vodafone account reminds me of the time when I have been trying to cancel the bank account at Česká spořitelna. Why does it have to be so hard to speak to a competent person in these big companies?
Mon, 05 Jun 2006
Apparently Novell did usability tests of their GNOME desktop (I think I have already read about it in the Linux Journal or somewhere). Today I have managed to browse their results: interesting work, I must say. However, there was one thing that surprised me, and I wondered whether the researchers did have any basic knowledge of how UNIX works at all:
In the "Set time and date"
test, the test subjects' task was to adjust the date and time on the local
system. One of the main problems was that they were confused when the
time setting application prompted for the root password - they incorrectly
supposed that they have to log out from the whole session and then login back
as a superuser. So far OK, I would say "fix the appearance of the
password dialog of
pam_timestamp (or whatever the responsible party is)".
However, the researchers' recommendation was: "Fix time and date settings to not require root access". WTF? This can be easily translated to "Make users log in as root by default", which is a behaviour of The Other OS, Which Should Not Be Named Here, and the behaviour which is often refered to as one of the main problems in the security of that OS. Bleeeh.
Fri, 02 Jun 2006
Some eight years ago, I wrote a driver for synchronous serial boards named COSA. COSA was an ISA-based board, so it is pretty obsolete these days. I occasionally fix the in-kernel driver, when somebody reports a bug (last fix was about half a year ago, IIRC). Today, an e-mail with the subject "COSA" arrived:
It was not a bug report, just some user of this driver sent me a mail, telling me that they are moving to a newer technology, and saying thanks for the driver, which worked for them for several years. The above photo (and two others) of their old Compaq Deskpro router with the COSA board (that green thing in the upper part) was attached. I was glad that my driver helped some people using flexible Linux-based routers instead of slow, expensive, and proprietary solutions (read: Cisco low-end routers). And of course, the driver helped the sales of COSA boards as well - I guess in the end, despite being originally designed for use with NetBSD, there might have been more Linux-based routers with COSA than NetBSD-based ones.
The insurance you don't need
I have a friend who likes to have everything insured, just in case something happens. He has even been trying to persuade me to buy an insurance for something, and he was offering to share his rebate with me. At that point I thought that from a basic fact of "the insurance company needs to make a profit" it is clear that every insurance is in fact (on average) a loss for the insured person.
So I think the insurance is only worth it, when you are not expected to have money to cover the damage yourself (such as when you buy a new car, and do not have money to immediately buy another one, when the first one is totaled or stolen). But for things like damages in order of low tens of thousands CZK (cca 30 CZK = 1 Euro), the insurance is acutally a nonsense, because you can well cover the damage yourself, without routing the money through the insurance company.
Today I have found an interesting article on K5: it has a title Don't buy insurance you don't need!, and the author gives my above thougths a more clear wording, and mentions also another things like deductible money. You may also want to read the comments, especially this one, which mentions the role of a bonus for claim-free time.