September 2006 Archives

Next Monday I'll be driving from Braga to Lisbon, for a symposium for all PhD students of Linguateca, next Tuesday and Wednesday. I have already the nights booked but in case someone wants to talk with me, please drop me an email.

I registered in 'last.fm'. This is my music profile:

I asked a friend about what this profile means. She answered: or you are gay, or you are a teenager girl. Oh well... :-|

Now that I know what my readers like (just see the number of comments on my calendar post), I can help those who like sexy photography to visit Keeley Hazell home page.

I need to buy this two-cd box. I am mostly a Queen fan, but I like Freedie Mercury solo a lot, too. Well, I can always order it from amazon together with the calendar (cough cough). Well, I need to think. At the moment, added it (the two-cd box) to my wish list at Amazon.

Yeah, I know, Procyon, I should not post this kind of stuff or people think I am pervert!

I really need to buy a calendar for 2007, and this seems a good choice. With a calendar like this one in my office at the University, I know my boss would go there make a visit more often.

Slackware, the linux distribution that I use, is preparing to release version 11. Thanks to Patrick Volkerding it continues alive, with good stability (sometimes, too much) and easy to use. Today I pre-ordered the 4 CDs case of Slackware. I think I really should to support the project.

A lot of people do not like Slackware, but I do not like the distributions that other people like, so, please, do not comment discussing how good or not Slackware is.

Finally we decided to port the Natura CVS repository to SVN. There is an application, named cvs2svn, which helps a lot on this task. Basically, it can convert all trunk revisions, and most branches. In some cases it fails but it is easy to find the problem and fix. Another good thing, is that it doesn't convert the repository in place. It creates a copy of the repository that you can remove entirely and try again in case it is not exactly what you want.

Also, for big projects repositories, it is not easy to know if you have everything that needs to be in the repository checked in. Thus, the cvs2svn application let you perform the conversion one module at a time, thus makes it easier to perform the transition.

Meanwhile, thanks to Strange and Krani1 for their help setting this up. I really hate Fedora Core but the owner of the machine like it. Given it was an old version (fc3) I needed to compile subversion, apache (and php). Also, I needed to compile them a lot of times. Oh, how I like Slackware.

Why in hell I teach (well, I play the first games) with someone who never played before, and after a few (no more than 10) he is winning all games? Oh well...

In the last days I've been programming a bigram/trigram/tetragram extractor. For those who do not know what this is, for the first sentence of this post, "In the", "the last", "last days" and so on are bigrams, "In the last" "the last days" are trigrams, and so on. As my program already mapped words into integers, I just need to store two, three or four integers, and the number of times they occurred by that same order.

My first implementation on my Mac, during a night in Zaragoza, used Berkeley DB 1. Berkeley DB 1 is builtin on Mac OS libc, thus making it easy to use it. The drawback was when trying to make it compile under Linux. DB support is not built-in under libc and just includes Berkeley DB 3 and Berkeley DB 4. The problem, one more time, is that Sleepycat should be sleeping, because they change the API every minor version. For instance, DB 4.1 and DB 4.2 have different APIs. Thus, it is not easy to write portable code using DB.

Second option (and current one, although I am not happy with it) is using SQLite 3. SQLite has the same advantage of DB (being standalone regarding a DBMS) and more, being relational makes it easier to write queries to bigrams, trigrams and tetragrams. But it has drawbacks, as well: for each bigram (or trigram or tetragram) that I find I need to first issue a "SELECT" statement to see if that bigram (or trigram or tetragram) was found previously and act accordingly: use a "UPDATE" to raise the occurrence number if it exists, or use a "INSERT" to create the entry.

For my small test corpus I have 17 89 168 bigrams (thus, this number of SELECTs, and the same number of UPDATEs or INSERTs) where 522 597 are unique bigrams. Also, 1 116 491 unique trigrams, and 1 391 123 tetragrams. These values just for one language.

My code using SQLite 3 is taking 30 minuts. Any ideas for faster solutions?

Yesterday come back from Zaragoza. That is a small city in northeast Spain, about 300Km from Barcelona. Its airport is basically a militar base with a small building for commercial flights. For what I could see, they have about 5 to 6 flights in the airpor dailly. Good for the airport shop, most of the flights are late.

The historic center of the city is quite small. In the other hand, outside the historic center there is not much to see. I walked a lot and just saw one McDonalds, two TelePizza and one Pans and Company. They have a fnac (Hurray) with a good selection of manga books (a lot of them, I must say) and a good selection of DVDs (not just movies, but a lot of documentaries, as well). The city has two 'El Corte Ingês', the typicall Spanish shopping center, but quite small when regarding to books, dvds, cds or toys.

While I could understand perfectly Spanish people, they couldn't understand me at all. I wonder if it is bad will or if it is really impossible for them to understand us.

More about spanish people, I met a chica mui guapa that I hope to meet again.

I just hope these little birds do not fall. I traveled before on Airbuses, but this is my first time in a Canadair...

And after Birmingham, next week is Zaragoza, Spain. It is much more near than Birmingham, but the truth is that I will need to take two flights as well. First, from Vigo to Madrid, and then from Madrid to Zaragoza. Both trips take one hour, more one hour in the airport. Also, I will need to go driving from home to vigo (more than one hour). That is not very bad. The really bad thing is that the coming back flight arrives at 11:00 pm at the Vigo airport. I hope not to be sleepy.

I just finished my first Email::Filter mail filter script. It is quite useful at the current moment, just the identify function. Now, send me spam and you will see what it will happen... muahahahaha

Today was movie day. Rebound was the choice. It is tagged as "comedy" on IMDB, so it is a good choice after somebody tell you some other person have died.

Oh, well, back to the movie. It is a typical movie. Good coach gets in troubles, need to coach a bad team (this case, stupid -- sorry boys -- kids), makes that team win, and refused to get back to the old job (thus, coaching again the kids). See? Typical. But it has good parts. Very good parts. It is funny. It has at lease one girl I can remember, but it is quite fun. OK, some of the jokes are well known, but most of them make you really laugh. See it!

Again, problems with credit cards. I do not own one. But, I am client of a service called MBnet, which creates virtual credit cards. I say the amount I need, and it issues a virtual credit card number. That credit card number works just fine, like any other credit card. The only problem is that it does not exist.

Why that is a problem?

I want to buy electronic tickets for my trip to a conference, next week, in Zaragoza. I visited the Iberia web site (the only company doing that trip) and asked for the ticket. Everything went fine until the end of the form, where it says: during your first check-in you will be asked for your credit card for identification.

I imagine that this is to restrict the use of robbed numbers... but... how can I show a virtual credit card??

What it means when a friend changes her photo in MSN to her belly-button? And a really sexy one? And yes, I asked her for the full photo but didn't have any luck... yet!

Unfortunately Portugal is HOT. Oh how I miss the fresh air from Birmingham.

I am sorry, but this is a fun game with some Portuguese friends, and thus, I will post this in Portuguese.

Seis coisas sobre mim:
1. Portista
2. Católico
3. Regente de um coro de Igreja
4. Toco piano mas não sei francês
5. Embora goste muito de uma menina ela não gosta de mim como eu dela
6. Sou programador inveterado

Seis pessoas a etiquetar:
1. UPS
2. Root (ficou sem blog, mas vai voltar a criar)
3. Luís Baixinho
4. Nocas
5. Rui Vilela
6. Rocha

I am in Birmingham, in Jury's Inn hotel. I asked for a non smoking room. I just do not know why am I smelling smoke (cigar smoke) right now.

Let me complain a little more about this hotel. In the first day they asked me for my credit card. I do not have one. Just a Visa Electron. Then, they asked to pay the first night. When I was preparing to pay the first night they say I need to pay all nights right in the first day. Of course I would need to pay it in the first day or in the last, but nowhere in the website they said it would be needed a credit card. Also, as you might know, normally you can't withdraw any value of money in an ATM. There is a daily limit (at least for Portuguese accounts). Thus, and as I had withdraw some before to pay the taxi, I couldn't withdraw the whole amount the hotel was asking me. Luckily enough, I was with a friend who had two different visa electron cards, from two different banks, and we could manage to get the total amount of money.

Now, just let me wait for tomorrow and see what else will happen when I do my check-out.

YAPC::EU::2006 is over. Nice conference, nice talks, unfortunately not a nice weather, and unfortunately one more time I think I do not know how to do properly presentations: presentations where people laugh and have a little fun.

Oh well, I'll try to get better next year.