Fossil Forum
danshearer
2 weeks, 1 day ago
Post: LumoSQL v0.82: self-hosting in Fossil
Hello all,
I just announced LumoSQL in the SQLite forum. From the point of view of Fossil/libfossil users:
- this version now works for all Fossil features as far as we know, with the sqlite.org repo and some larger ones. That includes the server and ui.
- encryption and page-based checksums work with LMDBv1.0, using a single file just like sqlite rather than the LMDB default of using a subdirectory
- the fossil/test directory contains correctness and performance scripts, and at the moment LumoSQL with LMDBv1.0 backend come back as faster on all measures
Fossil is an application and a pretty demanding one at that, so it is not surprising that the general performance graph has a pretty different shape to the Fossil one.
We have done a very Things Fossilers might want to do:
- try it!
- see if your repo or CI breaks LumoSQL-backed fossil
- try compiling on your favourite platform. We haven't done Windows or ARM, for example
- improve on our minor explorations of the use case of pathologically large repos
--
Dan Shearer
[email protected] / https://shearer.org
danshearer
2 weeks, 1 day ago
Hello all,
I just announced LumoSQL in the SQLite forum. From the point of view of Fossil/libfossil users:
- this version now works for all Fossil features as far as we know, with the sqlite.org repo and some larger ones. That includes the server and ui.
- encryption and page-based checksums work with LMDBv1.0, using a single file just like sqlite rather than the LMDB default of using a subdirectory
- the fossil/test directory contains correctness and performance scripts, and at the moment LumoSQL with LMDBv1.0 backend come back as faster on all measures
Fossil is an application and a pretty demanding one at that, so it is not surprising that the general performance graph has a pretty different shape to the Fossil one.
Things Fossilers might want to do:
- try it!
- see if your repo or CI breaks LumoSQL-backed fossil
- try compiling on your favourite platform. We haven't done Windows or ARM, for example
- improve on our minor explorations of the use case of pathologically large repos
--
Dan Shearer
[email protected] / https://shearer.org