Re: Ending work on snowball
Martin Porter <[email protected]> Wed, 12 Nov 2014 09:37:39 +0000
| Newsgroups | gmane.comp.search.snowball |
|---|---|
| Message-ID | <CALc6duwdD+Kiy7ybwwan+GOQMtgHdo2KTde8SH76X1DfRA=dig@mail.gmail.com> |
I think a github spin-off for future work is a good plan. The snowball.tartarus.org content I would like to see frozen in the immediate future. The texts I've written there, especially http://snowball.tartarus.org/texts/introduction.html , are "of their time", and will become increasingly out-of-date as time goes by. (It was all designed and written before the rise of Unicode as an international character standard.) But they should not be altered to catch up with current practice. The intro document is referenced in the literature, it was even (I learnt) reading material on an IR course, and it should remain in its current state. Similarly for the stemmers I've written. Having seen so many examples of Porter stemmer "improvements", I'd be alarmed at changes to the ones on snowball.tartarus.org which I had not authorised. Obviously a certain proprietorial jealousy on my part comes into play here. I hope it is excusable. Of course, there must be an explanation of the github development on the snowball site, with appropriate links (from the front page). The mailing list I've discussed with Richard at some length. We get a lot of spam, which Richard has been filtering out by hand. When he has not time to do this, a backlog builds up. I think it unfair Richard should have to do this. Many questions that come into the mailing list have been answered before, often many years ago. I have a private IR sytem for my emails that helps fish out old answers, but it is clumsy to relate it to the current email archive, I also get direct queries from people who can't bother to register with snowball mail. I am always dismayed when a question comes in on the mailing list which I can't answer: anything connected with snowball's java support, for example. I feel a webboard, with moderator, would work much better. We could keep the mailing list going, but I still think I would withdraw from it. My own feeling is that freezing the snowball site on 31 Dec 2014 is a good oppotunity for bringing to an end the current snowball mailing list.