History

Commits on git-commitgraph-v0.11.0 to experiments140+

test object-access read() + write() worst-case performance (#263)Sebastian Thiel committed 4 years agofa47115
Also actually upgrade the upgradable lock (#263)Sebastian Thiel committed 4 years ago25338c0
object-access tests for upgradable RwLock (#263)Sebastian Thiel committed 4 years ago658125a
Set object-access radicle-link experiment to latest master commitSebastian Thiel committed 4 years ago851af67
Add a warmup phase for object accessSebastian Thiel committed 4 years agof4e010e
object-access experiment with brand new radicle link Sync ODBSebastian Thiel committed 4 years ago7eb4168
Release git-features v0.18.0, git-actor v0.7.0, git-config v0.1.9, git-object v0.16.0, git-diff v0.12.0, git-traverse v0.11.0, git-pack v0.15.0, git-odb v0.25.0, git-packetline v0.12.2, git-transport v0.14.0, git-protocol v0.13.0, git-ref v0.10.0, git-repository v0.13.0, cargo-smart-release v0.7.0, safety bump 12 cratesSebastian Thiel committed 4 years agoacd3737
fmt (#263)Sebastian Thiel committed 4 years agofbeddeb
Adjust to chagne signature of peel_to_id_in_place (again?) (#263)Sebastian Thiel committed 4 years agodd5341f
change!: file::Store::(try_)find(…, packed) was removed (#263)Sebastian Thiel committed 4 years ago15d429b
thanks clippySebastian Thiel committed 4 years agodb1bb99
improve pack staleness handling (#259)Sebastian Thiel committed 4 years ago38e36a6
A good way to avoid using a potentially wrong pack (#259)Sebastian Thiel committed 4 years agoa14e4d0
realize that pack lookup needs the marker or wrong packs might be returned (#259)Sebastian Thiel committed 4 years agoa5a6a78
workaround borrow check, which in this case fits nicely (#259)Sebastian Thiel committed 4 years ago2977518
try loading actual packs with the current data structure… (#259)Sebastian Thiel committed 4 years ago87544c3
btree/hashmap free lookup of packs in store, keeping things more bundled (#259)Sebastian Thiel committed 4 years agoa88981b
Allow on-disk files to be reused as will in case they re-appear (#259)Sebastian Thiel committed 4 years ago6f5eaf1
refactorSebastian Thiel committed 4 years agod8d5a60
Simplify tracking of handles, using tokens to make that safer (#259)Sebastian Thiel committed 4 years ago407be6c
sketch a more concise way of keeping indices and packs relative to disk locations (#259)Sebastian Thiel committed 4 years ago6187b5c
Get a better understanding on how on-disk state can be represented in our state (#259)Sebastian Thiel committed 4 years ago58b0bcf
Learn about store's resource consumption (#259)Sebastian Thiel committed 4 years agoef9b08c
Make handles tracked to allow automatic handling of pack-unloading (#259)Sebastian Thiel committed 4 years ago70dc445
Flesh out Sync requirement for Repository (#259)Sebastian Thiel committed 4 years ago5180f11
A big step towards getting IDs straight, but… (#259)Sebastian Thiel committed 4 years ago5124abf
A better idea on how multi-pack indices (and changes to them) work (#259)Sebastian Thiel committed 4 years agofae3e7c
figure out how pack-ids can be stable even though multi-packs are still unclear (#259)Sebastian Thiel committed 4 years agobabfb7b
sketch a little more how packs could be accessed (#259)Sebastian Thiel committed 4 years ago3fce8f2
More work towards understanding indices and multi-pack indices for eventual pack lookup (#259)Sebastian Thiel committed 4 years agof90207d
Fix single-threaded version of design-sketch (#259)Sebastian Thiel committed 4 years ago3d533a1
Adjust unload mode to be a policy wide settings, probably along some more soon (#259)Sebastian Thiel committed 4 years ago4160b4a
RwLocks are preferred over Mutexes if there is a decent read-only path (#259)Sebastian Thiel committed 4 years ago9696383
Show that arcs can cost up to 16.5% when having 10mio ops compared to no-arc (#259)Sebastian Thiel committed 4 years agod40288a
This experiment shows that working through an Arc isn't slower than through a reference (#259)Sebastian Thiel committed 4 years ago977bd80