larsivi

As have already been noted on D.announce, I have started committing new code to DDBI. The task at hand is to refine the interfaces, make it easier to use where possible, and use as little resources as possible (this isn't necessarily easy given that fetched data need to stay in memory somewhere).

I have committed some new interfaces in dbi/model/ and implemented those (mostly at least) for mysql. SQLite will be the next to be added, and after that maybe PostgreSQL. I expect to have to rely on contributions for anything beyond that.

If you have any thoughts on the work I've been doing, please contact me. My current thoughts involve adding multi statement/multi result support (thanks BCS), make sure optional allocators can be used properly (and in all places applicable), how to best fetch results and error handling.

The first Tango Conference is approaching, and I'm giddy with expectations. I think the program is looking very good, and I'm happy that we got some speakers from across the pond.

The site of the conference is an excellent fit, with it being the first University to hold a D class. Torun, Poland is in addition a town with a very pretty medieval city centre, see the images here.

I expect to meet up with several of the other visitors already at the airport in Warsaw the day before the conference start, and except for Frank Benoit, these will be people I haven't met before, but that are active in the Tango community.

I'm to some degree active on LinkedIn and my answer to a question there led to the quotes that can be found in this article. Although the name isn't mentioned, my quotes are all taken from my work on Tango.

The article is about how easy/difficult it is to write/use open source software on Macs. Although I am fully behind my quotes there, she did leave out the ones that weren't so negative.

After the previous post on the topic, I have received quite a few followers on Twitter (15?) and 2 on Ohloh. I guess this means that I will use Twitter most, but I may duplicate what I consider important D/Tango messages to Ohloh, just to make it appear somewhat active there too (The Ohloh page takes its time updating the stats, so it gets kinda static without some other activity).

As for the Hackontest competition, they will start the selection process on August 1st. Apparently only 6 projects fulfill the requirements and are thus eligable. Tango isn't one of them, but is pretty close (missing one implementor).

Seems like it isn't all to long since last time I wrote about the D compilers - then about quality. This time I'd like to write about DMD's release process, as it is highly frustrating for me trying to have Tango compile with it.

Almost every release sees a regression, or some obscure accepts-invalid bug getting fixed in what is meant to be the stable branch. Today DMD 1.032 was released, with at least 3 issues causing immediate harm to Tango. One I made go away in SVN, but I'm considering reverting, and just calling 1.032 broken. The last 2 issues I don't know how they affect Tango yet. Actually, there may even be more that I just didn't get to yet.

I recently got my self a twitter account, and not very surprisingly you can find it at http://twitter.com/larsivi. The reasoning behind the move is that I don't want to post too much low quality content to this blog, and thus I end up seldomly really having to time to write here. With twitter I can make short notes and anecdotes without worrying about quality.

I'm not normally interested in the daily dealings of all people around the world, but see Twitter as an easy way to post thoughts around the open source work I do, and a bit of software development in general. Considering this, I think I find Ohloh's journals a better idea (or a better match towards what I want), but I'm not sure many enough of the D users use Ohloh to make the feature work for me. Also, Twitter could spread the "word" wider, although I guess the "word" would be considered less interesting to most potential readers.

I don't normally bother (or have time for) the various programming competitions on the net. This one caught my eye however, as I find the Tango implementation fairly short and elegant (the one that prompted my attention was in F# and wasn't particularly readable IMO, but then I'm not really into functional programming).

The task (see here to get your instance) is to process a directory tree (provided from Google in a zip file), and sum up values on certain lines of certain files.

My instance told me to sum values on line 5 in files with extension ".js" and with "BCD" somewhere in the path and multiply that with the sum of the values on line 1 in files with extension ".txt" and with "zzz" somewhere in the path. Empty lines should not be counted (I probably misunderstood something as an empty line would only have yielded a zero value for the sum in any case, but I got the correct result according to Google).

A few weeks ago, I entered Tango as a project for the Hackontest - a 24 hour hacking contest for 3 select features for various open source projects. The hacking will be done by teams of 3. For those attending, there will be cash prizes and most likely an exciting trip to Zurich, Switzerland.

Now, the catch is that a jury will select the 3 projects/features that will be part of the competition proper, and these will be chosen from those entered at the site linked further up.

When I registered Tango, there was a

A recurring complaint against D, is the quality of the compilers. Currently there are two in a usable state - DMD and GDC, with LLVMDC, Dil and Dang as follow ups.

This post is about the first two, as I don't consider a D compiler usable until it can compile Tango and its examples.

DMD is quite stable, especially its 1.0x branch - but the most annoying bugs - those that it is hardest to find workarounds for - tend to have a low priority. The reasoning seems mostly to be that the fruits are hanging to high. Also, it does have fairly unstable optimization - while developing Tango XML, just moving a function in the source could

KeYeR (Piotr) called upon me in #D and said that my statement "This is one of the worse decisions among the bad ones in the D history." was bad English. No, not really. He said that it was a strong statement, and sure, it is. I tend to be (unnecessarily so?) strong in my rather few statements on design choices in the D language. Peter added that he was afraid that I was right.

As I see it, Walter here is willingly implementing a solution that 99% of the community seems to hate. He even had a different implementation, the manifest keyword, that was applauded