1 month, 2 weeks ago
star star star_half star_border star_border
Mostly interesting, but a little strange in two ways.
The authors are the same pair who wrote the BOHM paper, but in this overview they never identify themselves as such despite referring directly to the BOHM paper multiple times.
Second, up until the last page the article exclusively extolls the virtues of deterministic DBs, explaining how they improve on non-deterministic systems or how their shortcomings are more than adequately offset by the benefits. However, in the last section they briefly mention two major downsides: required input preprocessing and lack of support for interactive transactions. Then they conclude by stating that deterministic DBs are unusable until these two problems are resolved. The conclusion strikes me as a bizarre non-sequitur given the preceding 13 pages. Even still, I have to credit the authors for the self-awareness in clearly articulating the reasons for their system's non-adoption, 3 years after they published the BOHM paper.