Some old university home directories contain class notes and a photograph of somebody’s cat. Oleg Bartunov’s ~megera directory became one of the working addresses attached to major pieces of PostgreSQL development.
Visit Oleg Bartunov’s old SAI web directory
Bartunov worked at the Sternberg Astronomical Institute of Moscow State University and has been involved with PostgreSQL since the 1990s. Conference biographies credit him with introducing locale support and, with Teodor Sigaev, developing or advancing infrastructure including GiST, GIN, full-text search and extensions such as hstore, pg_trgm, ltree and others.
A surprising amount of that history still points back into the same university URL tree.
The surviving pages include development notes for GiST and GIN indexes, tsearch2 documentation, full-text-search experiments, trigram matching, hstore, test data, presentations, benchmarks, TODO lists and utilities intended for PostgreSQL developers.
One old pg_trgm README, for example, describes trigram similarity searching and explicitly points readers to the Tsearch2 and GiST development sites under the ~megera directory. PostgreSQL mailing-list posts did the same thing. In 2010 Bartunov gave other developers a URL there for a 580 MB test dataset so they could reproduce his GIN-index results.
This was not a brochure about software development.
It was software development happening on the Web.
The research homepage became project infrastructure
Bartunov’s database work grew out of astronomy. Later PostgreSQL conference biographies explain that while working at Sternberg Astronomical Institute he needed a capable open-source database for scientific work. That eventually connected astronomical data problems with database indexing research.
The result is a particularly good example of how open-source projects used the old Web. Documentation, experiments, tarballs, talks, benchmarks and mailing-list conversations did not necessarily live in one polished project portal. They lived wherever the people doing the work had an account and some disk space.
The root of the old site is less reliable today than some of its deep pages, but substantial material under the directory still survives and is still indexed. More importantly, contemporary PostgreSQL mailing-list archives preserve the links and show how developers actually used them.
For software historians, that context matters. Modern documentation can tell you what a feature does now. An old development page can show what the authors were testing before the feature settled into its present form.
CacheRat’s 1,967 Ancient Web Domains research list includes technical personal and university sites because open-source history is scattered through exactly these kinds of directories.
A tilde account at an astronomy institute helped document database features used around the world.
That is a pretty good return on some university web space.