Recent blog entries for gobry

Just did something painful: decided to potentially kill an open-source project I've been working on for some years now. Not enough time and the fact that I'm not a user of my own project anymore: I've been trying to introduce new features from time to time, mostly things I fancied. But the fact is: if you don't scratch your own itches, it's hard to put in enough efforts to turn a proof of concept into a product.

gdvieira: it makes sense to me to have c+' = ć and c+, = ç (which is what I get from the apps I've tested 1 minute ago)

Just started reading the D Programming Language website. Looks too good to be true: efficient language with high level constructs, native support for unittests and design by contracts, direct access to native C APIs... Has anybody here used it for something larger than a toy project, and would like to share his experience?

gilbou: it looks like whois is doing a wildcard search, which returns these legitimate but dumb entries....

Pfewww, I really pushed hard to get a new version of Pybliographer out of the door. I'm still amazed at how much one can do alone when using the proper tools. In this case, the proper tool is Python.

Context: my pet project is a bibliography management tool. Its purpose is to help you organise references, search for new material on external databases, help you cite these references in documents,... I've been able to integrate my code with OpenOffice (thanks PyUNO), with LyX (I won't thanks unix pipes on this one), define a proper citation mini-language, put a GUI around part of it (thanks PyGTK) and test some parts of it (thanks, dogtail), send queries to PubMed, Web of Science, CrossRef (thanks Twisted), parse their results (thanks ElementTree), and even package the whole easily (thanks setuptools)!

A while ago, I said that the purpose of pyblio was not to replace tools like EndNote and the like. Maybe I should revise this. Areas where I'd like to push involve reference sharing and fulltext management. I've already a few ideas, but I need to see if the recent features help the project get more traction.

Thanks Raph, and thanks to all of you who posted so many interesting stuff here. It really has been a pleasure to read advogato.

(thanks to the contributors that brought my diary score almost at 3 :-))

Instead of simply documenting how to use pyblio, I've started working on simple demo code that actually use the framework, and are also immediately useful. The first one is an online PubMed to BibTeX converter. Maybe this will help people start using it.

(hello? don't know if anybody reads this, I'm the first to admit I don't write especially interesting stuff, but c'mon I don't think my entries deserve a score as low as our friendly chinese spammers, do I? so if anybody with a heart would be so nice and give me at least a score of 2??)

Pyblio has gained a nice feature. For the record, it's a python framework for handling publication references, but probably also your collection of books, dvds, tapes, whatever. In pyblio 1.3, you can now adapt a publication database format (from, say, PubMed) into another format (say, BibTeX). It's still more of a proof of concept, waiting for more extensive use. So far I have the following features:

• external queries (pubmed, web of science, crossref)
• adaptation between database formats
• citation formatting (to generate an actual citation from a publication reference)
• searching / editing / storing of references

What I need to complete is:

• storage of fulltext documents (I've a few ideas already)
• interaction with word processors (esp. openoffice)
• glueing all these features in the GUI, which is very rudimentary so far on the 1.3 branch.

I would like to extend the data model a bit, but this will certainly wait until the rest consolidates.

Unfortunately, the community that lives around pyblio has a low ratio of developers / users. Worse, as I had not enough free time these last months for the project, I probably discouraged some potential developers. I really need to lower the barrier of entry into the project, but I don't know how so far. Perhaps by completing a few sample applications instead of adding new features? Writing documentation? Putting some simple web apps online, so that people can actually test some features?

wnewman: having only experimented on toy projects with haskell and common lisp, I don't have a very deep insight, but I had in both cases the feeling that the downsides (hard to predict runtime behavior for haskell, lack of some important common blocks like standard sockets for common lisp) were like a cold shower after you make the initial step of learning these languages and discover their bright sides (and they really shine in many aspects).

python position in switzerland

I'm leaving my current job at the end of june, and this will make a nice open position for a hacker wishing to work on a open-source project written in python, in a very nice area of Switzerland (french-speaking area BTW).

The job is at EPFL (Swiss Institute of Technology), and is about continuing the development of a service called infoscience which provides researchers of more than 200 labs with a common platform for managing their publications. It is not (only) an archive, as it behaves as a web service for plenty of web sites, and also pumps information from external systems.

So basically the job is about balancing the demand for specific tools (for a given lab, for the librarians,...) and the need for a high-level common platform (which already provides conversion services, quality checking tools,...) Part of the software is developed in collaboration with the CDS team at CERN, part is specific to infoscience and its design is completely up to the person in charge of this job.

The detailed offer and the procedure for applying is available on the EPFL jobs website, in french. Feel free to contact me directly for more info.

(and no, I don't leave because it's boring, but because I've been recruited by Google :-))