Month: July 2007

Do your laptop fans produce a lot of noise?

Someone hoped my laptop doesn’t make too much noise after I posted a photo of the Tecra logo on Flickr. The short answer is no, it doesn’t make too much noise. At 10cm from the fan output, I can measure 42dB when the fan is off and 52dB when it’s on.

Beside the fact that I don’t hear that noise when I have my headphones, it was not sufficient for me. I wrote small python and gnuplot scripts to collect and display temperature, fan status and load (.tar.gz file, 1.3ko). During those 2 hours, I checked my e-mails, read news on the web and wrote the OPML output in catrss (that’s why load averages increase at the end, when I’m debugging the software). Here are the results (click on an image to see a larger version):

Graph of temperature and load on a Toshiba Tecra S1 for 2 hours after boot

Graph of temperature and fan activity on a Toshiba Tecra S1 for 2 hours after boot (see text for explanation on fan activity)

One strange thing is that the status of both fans (/proc/acpi/fan/FAN0/state and /proc/acpi/fan/FAN1/state) is always off. But, when you ask for the trip points (/proc/acpi/thermal_zone/THZN/trip_points), you see that FAN1 should be active above 45Â°C and FAN0 should be active above 104Â°C! Practically, after some observations, I realized that FAN1 is activated if the temperature is equal to or above 50Â°C and it doesn’t stop until temperature is equal to or below 45Â°C. That’s the behaviour displayed in green on the second chart.

OPML output in catrss

A few days ago, I released the first version of catrss, a tool used to concatenate RSS file(s) to standard output. Today, I added OPML output to this tool. Here it is in version 0.2 (.tar.gz file, 16ko).

OPML is a file format first used in a commercial application. Now it’s widely used for the exchange of links between news aggregators. Because of that, I had to implement it in catrss: it’s a potential format for the output of catrss.

But I’m not happy with this format. It’s nice, it does its job, it’s already available, … ok. Other people already complained about this format (see here e.g.). Beside the fact it’s not standardized (it’s difficult to exactly know what software will correctly parse what you put in your outline tags), I dislike the fact that the official specifications include tags that describe what I consider as the look, feel and behaviour of the content (expansionState, vertScrollState and the window*) and not the content itself. Of course, I’ve read that they are optional. Nevertheless, I don’t think it’s a good design decision to handle things like this and mix look/behaviour and content (not in this case).

I’m sorry for people not in computer science but this is — again — another entry about programming and I understand the title and content are a bit strange for them.

Picklist Editor 0.2

I’ve just released the version 0.2 of Picklist Editor. Now you have a table of all the proteins on the right of the gel. If you double-click on a cell, you can edit it (note this is not a recommended behaviour). After revalidating the table, your new spot will be included in the gel (and saved to your picklist if you like it). For me, this version is stable and fully functional 🙂

Picklist Editor 0.2 screenshot

The hardware side of Picklist Editor 0.1

This morning, I released Picklist Editor 0.1 with a text introduction … Hmmm … on my photos on Flickr you can see the hardware side of the picking process … (click on pictures to see details).

On the photo on the left, you can see a gel on a low-fluorescent glass plate. This plate is in part in a tray that firmly holds it when the robot is doing its job. The holes everywhere result from the picking process but there are proteins everywhere and you can’t see them in visible light since they are labelled with fluorescent Cy dyes. You can see two white round stickers on each side of the gel: these are the picking references.

Here, the picker head is in the process of taking a part of the gel with some proteins inside. Exact positions were computed according to the fluorescent images, revealing the proteins. As you can see again: the gel is perfectly transparent for our non-bionic eyes.

Finally, you can see the spot picking robot in action. The picking head is moving following two axis thanks to the horizontal bar at the back and the perpendicular arm holding the picking head and camera. On the left, you have a pumping station: in addition to some jazz when the picking head is on the gel, the station is aspiring water through the head in order to help getting a plug out of the gel. After that the arm moves to the right of the photo where you have two 96-wells plates to collect samples. When the head is above a well, the pumping station is “blowing” water into the head in order to eject the plug into the well. Everything is under control of a computer and software that is on the right, outside of the camera angle.

This is the GE Healthcare “Typhoon 9400” scanner used to scan fluorescent gels. It’s a huge beast but it doesn’t make a lot of noise (well, I don’t want to stay the whole day next to it!). And this unit only has the red and green lasers inside. There is a second (smaller) unit below with only a blue laser source in it! You can see two gels ready to be scanned (the upper door has to be closed before!).

Next is the analysis “workstation” where images of the gels are analyzed (after scanning and before spot picking). The software (DeCyder) helps to create pick lists.

Other photos from my labs can be seen in my laboratory photostream.

Picklist Editor 0.1

When you work with 2D gel electrophoresis in proteomics, you end dealing with "pick lists". For this purpose, I wrote Picklist Editor, a tool to help visualize and modify this pick list.

Picklist Editor 0.1 screenshot

As usual, software and source code are available here. Feel free to use it and report any bug or your wish list 🙂

(and if you didn’t understand everything above because you are not in the proteomics field, just go to the page too because I also wrote a small introduction)

I should change this blog title to “Version 0.1” since I recently released version 0.1 of many software tools 😉

catrss 0.1

One day, one has to sit at his/her table and try to really understand how to deal with XML. Since I think I can only learn with a project in mind, I took Alexandre Dulaunoy’s mergerss suggestion and tried to develop my own catrss.

As the name implies, catrss is one of the many descendants of the cat command. Catrss is used to concatenate RSS file(s) to standard output. In its most simple form, you simply have to give it some RSS files to parse and it will concatenate them for you ; the command is:

./catrss rssfile1.xml rssfile2.xml ...

If you want to see all the parameters you can set, just type “./catrss –help”. You’ll probably prefer to set your own title, link and description parameters since they are the only mandatory elements. One important point to keep in mind is that, by default, catrss only take the 10 most recent items (blog entries, e.g.) from all the files. You can change this value with the “-n” option.

For the moment, catrss is only available here (.tar.gz file, 16ko). The file contains the catrss program, its source code and two example of RSS files. Code is licensed under the GNU GPL. You need only Python 2.5 in order to run catrss (it’s probably already installed on any GNU/Linux computer).

Currently, it only works with RSS 2.0 files and it’s very picky with dates (for example, it’s not working with this blog RSS stream — what a shame!). But all this could be improved for version 0.2. Suggestions, bug reports and patches are welcome.

Finally, dealing with XML and Python is very easy. ElementTree documentation is quite good. And, except for other Unix-minded tools, there is plenty of other cool stuff one can do with XML: parse answers from the Yahoo API, deal with XML-RPC and other web services, …

Of course, it’s when you are struggling to feed XML into your program that you realize other people already developed what you are just doing: I’ve found at least 5 RSS parsers/generators [1, 2, 3, 4, 5] and 3 tutorials [1, 2, 3]. But I’m proud to say I didn’t used any of these references for catrss.

oncolour – a Flickr add-on for background

After discovering the Flickr API, I started coding oncolour this night and here is the result … oncolour is a PHP script that allows you to display your photos from Flickr on your own website and with a specific background colour. It’s better with an example … This URL "http://www.epot.org/flickr/oncolour.php?id=860125589," will give this result:

oncolour screenshot

Some other solutions have been developed and used (example) but this one is really free:

free for use (see here how to use it – even for your photos – and a description of all the options) and
free to re-use (see here how to download the script and use it on your own server).

Feel free to use it! 🙂

GeoPolis – watching the watchdogs

Quuxlabs is releasing the alpha version of GeoPolis, a web service to gather and show the ongoing police control. More info can be found on the project website and on the Quuxlabs blog. I made some comments regarding the service accuracy, automation, security and independence but I’ll be glad to use the service once I’ll be able to drive again.

Btw, in the same topic — watching the watchdogs, my map of cameras in Liege (see announcement and update) is now located on the CCLV website. The database management is also handed over to them. Moreover Christophe Cattelain is now systematically taking photos of those cameras in Liege and he publishes them on Flickr …

Getting some TV news programmes

I told you it’s boring to lay down the whole day (see previous post). And even if I have a laptop, it’s very uncomfortable to type when you are on your bed with a leg on top of 3 pillows. Anyway, I’m not here to talk about my life but to share two small Python scripts. Their goal is to retrieve two television evening news programmes (from RTBF1 and France2, both in French). With that, I can directly watch evening news from my laptop (no need to browse their website nor install ad hoc Firefox plugin; everything can be done from the command line).

For the RTBF1 evening news, it was quite easy since they just rely on a 3rd-party hosting company to provide the video stream. One has just to find the correct URL and voilà: getRTBFvideo.py (446b).

For the France2 evening news, it was a little bit more tricky since they give the programme date in the URL. In my script, it gives yesterday URL if it’s launched before 9.00pm (their programme starts at 8.00pm every evening so it’s reasonable not to download today programme before 9.00pm). Here is the script: getF2video.py (889b).

Both scripts only give you the URL of the video stream. You need then to feed this URL into your favorite multimedia player (mplayer, e.g.). Please note you’ll still need proprietary codecs in order to view these streams.

Miro …

It’s so boring to lay down the whole day (sprain at left ankle), I decided to give a try to Miro, a software to “watch internet video channels as easy as to watch TV”. Before trying it, I tried to watch their video demonstration but it’s not working since it requires the proprietary Flash player …

Miro video demonstration not working without Flash

Now, if you want to try Miro, do not use “yum install Democracy” on Fedora 7 (Democracy Player is Miro former name). It will give you version 0.9.5.1-10 that will cause many problem (the main one being it won’t run).

If you want to try Miro, go to their download page and download the ad hoc package (i.e. version 0.9.8-1 when I’m writing this). Try to install it by typing “rpm -i Miro…” and resolve the dependencies (I had to download libfame). If, like me, you are already using Firefox 2.0.0.4, it will still require version 2.0.0.3. You can force the installation by typing “rpm -i –nodeps Miro…”. And it still doesn’t work. I get some errors apparently linked with dbus. I posted a comment on the Miro forum. I hope to get an answer soon …

Meanwhile, it’s still boring to lay down the whole day …