The first issue was the file structure. For the other files in the TMER series -- the most recent on this blog was Karpov TMER (1985-) -- I created one PGN file per event for the player. Then I bundled them all into a single ZIP file. This simplifies maintenance, but takes an additional step to load all of the games into a single database. I can do this easily, but other people might not have the tools or know the technique. For this first release, I kept all of the games in a single PGN file.
The second issue was how much data to carry in the PGN headers ('[Event "Candidates match"]' etc.). I'm a minimalist, but the modern tendency is to load every bit of known data about a game into its headers. Although this keeps everything together, the cost is huge files that are more 'machine friendly' than 'people friendly'. An XML-style approach would be more appropriate and I should investigate what has been done for chess software. In the end I decided to go with the basic seven-tag standard roster plus an ECO tag.
While I was doing this I encountered some format issues in the game scores when my software complained about some strange use of the '#' (checkmate) character. I left the problems in the file and will come back to them later.
The PGN file, along with a text index suitable for a database, is available at carlsen.zip. I'll also add a pointer to my first post on the subject, A Magnus Carlsen Game Collection.
(TMER = Tournament, Match, and Exhibition Record)
No comments:
Post a Comment