Mercurial > hg
view tests/test-newbranch.t @ 40326:fed697fa1734
sqlitestore: file storage backend using SQLite
This commit provides an extension which uses SQLite to store file
data (as opposed to revlogs).
As the inline documentation describes, there are still several
aspects to the extension that are incomplete. But it's a start.
The extension does support basic clone, checkout, and commit
workflows, which makes it suitable for simple use cases.
One notable missing feature is support for "bundlerepos." This is
probably responsible for the most test failures when the extension
is activated as part of the test suite.
All revision data is stored in SQLite. Data is stored as zstd
compressed chunks (default if zstd is available), zlib compressed
chunks (default if zstd is not available), or raw chunks (if
configured or if a compressed delta is not smaller than the raw
delta). This makes things very similar to revlogs.
Unlike revlogs, the extension doesn't yet enforce a limit on delta
chain length. This is an obvious limitation and should be addressed.
This is somewhat mitigated by the use of zstd, which is much faster
than zlib to decompress.
There is a dedicated table for storing deltas. Deltas are stored
by the SHA-1 hash of their uncompressed content. The "fileindex" table
has columns that reference the delta for each revision and the base
delta that delta should be applied against. A recursive SQL query
is used to resolve the delta chain along with the delta data.
By storing deltas by hash, we are able to de-duplicate delta storage!
With revlogs, the same deltas in different revlogs would result in
duplicate storage of that delta. In this scheme, inserting the
duplicate delta is a no-op and delta chains simply reference the
existing delta.
When initially implementing this extension, I did not have
content-indexed deltas and deltas could be duplicated across files
(just like revlogs). When I implemented content-indexed deltas, the
size of the SQLite database for a full clone of mozilla-unified
dropped:
before: 2,554,261,504 bytes
after: 2,488,754,176 bytes
Surprisingly, this is still larger than the bytes size of revlog
files:
revlog files: 2,104,861,230 bytes
du -b: 2,254,381,614
I would have expected storage to be smaller since we're not limiting
delta chain length and since we're using zstd instead of zlib. I
suspect the SQLite indexes and per-column overhead account for the
bulk of the differences. (Keep in mind that revlog uses a 64-byte
packed struct for revision index data and deltas are stored without
padding. Aside from the 12 unused bytes in the 32 byte node field,
revlogs are pretty efficient.) Another source of overhead is file
name storage. With revlogs, file names are stored in the filesystem.
But with SQLite, we need to store file names in the database. This is
roughly equivalent to the size of the fncache file, which for the
mozilla-unified repository is ~34MB.
Since the SQLite database isn't append-only and since delta chains
can reference any delta, this opens some interesting possibilities.
For example, we could store deltas in reverse, such that fulltexts
are stored for newer revisions and deltas are applied to reconstruct
older revisions. This is likely a more optimal storage strategy for
version control, as new data tends to be more frequently accessed
than old data. We would obviously need wire protocol support for
transferring revision data from newest to oldest. And we would
probably need some kind of mechanism for "re-encoding" stores. But
it should be doable.
This extension is very much experimental quality. There are a handful
of features that don't work. It probably isn't suitable for day-to-day
use. But it could be used in limited cases (e.g. read-only checkouts
like in CI). And it is also a good proving ground for alternate
storage backends. As we continue to define interfaces for all things
storage, it will be useful to have a viable alternate storage backend
to see how things shake out in practice.
test-storage.py passes on Python 2 and introduces no new test failures on
Python 3. Having the storage-level unit tests has proved to be insanely
useful when developing this extension. Those tests caught numerous bugs
during development and I'm convinced this style of testing is the way
forward for ensuring alternate storage backends work as intended. Of
course, test coverage isn't close to what it needs to be. But it is
a start. And what coverage we have gives me confidence that basic store
functionality is implemented properly.
Differential Revision: https://phab.mercurial-scm.org/D4928
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Tue, 09 Oct 2018 08:50:13 -0700 |
parents | 1c7167009936 |
children | 8197b395710e |
line wrap: on
line source
$ branchcache=.hg/cache/branch2 $ listbranchcaches() { > for f in .hg/cache/branch2*; > do echo === $f ===; > cat $f; > done; > } $ purgebranchcaches() { > rm .hg/cache/branch2* > } $ hg init t $ cd t $ hg branches $ echo foo > a $ hg add a $ hg ci -m "initial" $ hg branch foo marked working directory as branch foo (branches are permanent and global, did you want a bookmark?) $ hg branch foo $ hg ci -m "add branch name" $ hg branch bar marked working directory as branch bar $ hg ci -m "change branch name" Branch shadowing: $ hg branch default abort: a branch of the same name already exists (use 'hg update' to switch to it) [255] $ hg branch -f default marked working directory as branch default $ hg ci -m "clear branch name" created new head There should be only one default branch head $ hg heads . changeset: 3:1c28f494dae6 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: clear branch name Merging and branches $ hg co foo 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg branch foo set existing branch name fails unless force - setting existing parent branch works without force: $ hg branch bar abort: a branch of the same name already exists (use 'hg update' to switch to it) [255] $ hg branch -f bar marked working directory as branch bar $ hg branch foo marked working directory as branch foo $ echo bleah > a $ hg ci -m "modify a branch" $ hg merge default 0 files updated, 0 files merged, 0 files removed, 0 files unresolved (branch merge, don't forget to commit) $ hg branch foo set existing branch name where branch head is ancestor: $ hg branch bar abort: a branch of the same name already exists (use 'hg update' to switch to it) [255] set (other) parent branch as branch name $ hg branch default marked working directory as branch default set (first) parent branch as branch name $ hg branch foo marked working directory as branch foo $ hg ci -m "merge" $ hg log -G -T '{rev}:{node|short} {branch} {desc}\n' @ 5:530046499edf foo merge |\ | o 4:adf1a74a7f7b foo modify a branch | | o | 3:1c28f494dae6 default clear branch name | | o | 2:c21617b13b22 bar change branch name |/ o 1:6c0e42da283a foo add branch name | o 0:db01e8ea3388 default initial $ hg branches foo 5:530046499edf default 3:1c28f494dae6 (inactive) bar 2:c21617b13b22 (inactive) $ hg branches -q foo default bar Test for invalid branch cache: $ hg rollback repository tip rolled back to revision 4 (undo commit) working directory now based on revisions 4 and 3 $ cp ${branchcache}-served .hg/bc-invalid $ hg log -r foo changeset: 4:adf1a74a7f7b branch: foo tag: tip parent: 1:6c0e42da283a user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: modify a branch $ cp .hg/bc-invalid $branchcache $ hg --debug log -r foo changeset: 4:adf1a74a7f7b4cd193d12992f5d0d6a004ed21d6 branch: foo tag: tip phase: draft parent: 1:6c0e42da283a56b5edc5b4fadb491365ec7f5fa8 parent: -1:0000000000000000000000000000000000000000 manifest: 1:8c342a37dfba0b3d3ce073562a00d8a813c54ffe user: test date: Thu Jan 01 00:00:00 1970 +0000 files: a extra: branch=foo description: modify a branch $ purgebranchcaches $ echo corrupted > $branchcache $ hg log -qr foo 4:adf1a74a7f7b $ listbranchcaches === .hg/cache/branch2 === corrupted === .hg/cache/branch2-served === adf1a74a7f7b4cd193d12992f5d0d6a004ed21d6 4 c21617b13b220988e7a2e26290fbe4325ffa7139 o bar 1c28f494dae69a2f8fc815059d257eccf3fcfe75 o default adf1a74a7f7b4cd193d12992f5d0d6a004ed21d6 o foo Push should update the branch cache: $ hg init ../target Pushing just rev 0: $ hg push -qr 0 ../target $ (cd ../target/; listbranchcaches) === .hg/cache/branch2-base === db01e8ea3388fd3c7c94e1436ea2bd6a53d581c5 0 db01e8ea3388fd3c7c94e1436ea2bd6a53d581c5 o default Pushing everything: $ hg push -qf ../target $ (cd ../target/; listbranchcaches) === .hg/cache/branch2-base === adf1a74a7f7b4cd193d12992f5d0d6a004ed21d6 4 c21617b13b220988e7a2e26290fbe4325ffa7139 o bar 1c28f494dae69a2f8fc815059d257eccf3fcfe75 o default adf1a74a7f7b4cd193d12992f5d0d6a004ed21d6 o foo Update with no arguments: tipmost revision of the current branch: $ hg up -q -C 0 $ hg up -q $ hg id 1c28f494dae6 $ hg up -q 1 $ hg up -q $ hg id adf1a74a7f7b (foo) tip $ hg branch foobar marked working directory as branch foobar $ hg up 0 files updated, 0 files merged, 0 files removed, 0 files unresolved Fast-forward merge: $ hg branch ff marked working directory as branch ff $ echo ff > ff $ hg ci -Am'fast forward' adding ff $ hg up foo 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ hg merge ff 1 files updated, 0 files merged, 0 files removed, 0 files unresolved (branch merge, don't forget to commit) $ hg branch foo $ hg commit -m'Merge ff into foo' $ hg parents changeset: 6:185ffbfefa30 branch: foo tag: tip parent: 4:adf1a74a7f7b parent: 5:1a3c27dc5e11 user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: Merge ff into foo $ hg manifest a ff Test merging, add 3 default heads and one test head: $ cd .. $ hg init merges $ cd merges $ echo a > a $ hg ci -Ama adding a $ echo b > b $ hg ci -Amb adding b $ hg up 0 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ echo c > c $ hg ci -Amc adding c created new head $ hg up 0 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ echo d > d $ hg ci -Amd adding d created new head $ hg up 0 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ hg branch test marked working directory as branch test (branches are permanent and global, did you want a bookmark?) $ echo e >> e $ hg ci -Ame adding e $ hg log changeset: 4:3a1e01ed1df4 branch: test tag: tip parent: 0:cb9a9f314b8b user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: e changeset: 3:980f7dc84c29 parent: 0:cb9a9f314b8b user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: d changeset: 2:d36c0562f908 parent: 0:cb9a9f314b8b user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: c changeset: 1:d2ae7f538514 user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: b changeset: 0:cb9a9f314b8b user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: a Implicit merge with test branch as parent: $ hg merge abort: branch 'test' has one head - please merge with an explicit rev (run 'hg heads' to see all heads) [255] $ hg up -C default 1 files updated, 0 files merged, 1 files removed, 0 files unresolved Implicit merge with default branch as parent: $ hg merge abort: branch 'default' has 3 heads - please merge with an explicit rev (run 'hg heads .' to see heads) [255] 3 branch heads, explicit merge required: $ hg merge 2 1 files updated, 0 files merged, 0 files removed, 0 files unresolved (branch merge, don't forget to commit) $ hg ci -m merge 2 branch heads, implicit merge works: $ hg merge 1 files updated, 0 files merged, 0 files removed, 0 files unresolved (branch merge, don't forget to commit) $ cd .. We expect that bare update on new branch, updates to parent $ hg init bareupdateonnewbranch $ cd bareupdateonnewbranch $ hg update 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ touch a $ hg commit -A -m "a" adding a $ touch b $ hg commit -A -m "b" adding b $ touch c $ hg commit -A -m "c" adding c $ hg update -r 1 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ hg log -G o changeset: 2:991a3460af53 | tag: tip | user: test | date: Thu Jan 01 00:00:00 1970 +0000 | summary: c | @ changeset: 1:0e067c57feba | user: test | date: Thu Jan 01 00:00:00 1970 +0000 | summary: b | o changeset: 0:3903775176ed user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: a $ hg branch dev marked working directory as branch dev (branches are permanent and global, did you want a bookmark?) $ hg update 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg summary parent: 1:0e067c57feba b branch: dev commit: (new branch) update: (current) phases: 3 draft $ cd .. We need special handling for repositories with no "default" branch because "null" revision belongs to non-existent "default" branch. $ hg init nodefault $ cd nodefault $ hg branch -q foo $ touch 0 $ hg ci -Aqm0 $ touch 1 $ hg ci -Aqm1 $ hg update -qr0 $ hg branch -q bar $ touch 2 $ hg ci -Aqm2 $ hg update -qr0 $ hg branch -q baz $ touch 3 $ hg ci -Aqm3 $ hg ci --close-branch -m 'close baz' $ hg update -q null $ hg log -GT'{rev} {branch}\n' _ 4 baz | o 3 baz | | o 2 bar |/ | o 1 foo |/ o 0 foo a) updating from "null" should bring us to the tip-most branch head as there is no "default" branch: $ hg update -q null $ hg id -bn -1 default $ hg update 2 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -bn 2 bar b) but if we are at uncommitted "default" branch, we should stick to the current revision: $ hg update -q 0 $ hg branch default marked working directory as branch default $ hg id -bn 0 default $ hg update 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -bn 0 default c) also, if we have uncommitted branch at "null", we should stick to it: $ hg update -q null $ hg branch new marked working directory as branch new $ hg id -bn -1 new $ hg update 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -bn -1 new $ cd .. We expect that update --clean discard changes in working directory, and updates to the head of parent branch. $ hg init updatebareclean $ cd updatebareclean $ hg update --clean 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ touch a $ hg commit -A -m "a" adding a $ touch b $ hg commit -A -m "b" adding b $ touch c $ hg commit -A -m "c" adding c $ hg log changeset: 2:991a3460af53 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: c changeset: 1:0e067c57feba user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: b changeset: 0:3903775176ed user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: a $ hg update -r 1 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ hg branch new-branch marked working directory as branch new-branch (branches are permanent and global, did you want a bookmark?) $ echo "aa" >> a $ hg update --clean 2 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg status $ hg branch default $ hg parent changeset: 2:991a3460af53 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 summary: c We expect that update --clean on non existing parent discards a new branch and updates to the tipmost non-closed branch head $ hg update null 0 files updated, 0 files merged, 3 files removed, 0 files unresolved $ hg branch newbranch marked working directory as branch newbranch (branches are permanent and global, did you want a bookmark?) $ hg update -C 3 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg summary parent: 2:991a3460af53 tip c branch: default commit: (clean) update: (current) phases: 3 draft