Mercurial > hg
view tests/test-subrepo-svn.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 | 5495ceab5637 |
children | eb6700e6c5ea |
line wrap: on
line source
#require svn15 $ SVNREPOPATH=`pwd`/svn-repo #if windows $ SVNREPOURL=file:///`"$PYTHON" -c "import urllib, sys; sys.stdout.write(urllib.quote(sys.argv[1]))" "$SVNREPOPATH"` #else $ SVNREPOURL=file://`"$PYTHON" -c "import urllib, sys; sys.stdout.write(urllib.quote(sys.argv[1]))" "$SVNREPOPATH"` #endif $ filter_svn_output () { > egrep -v 'Committing|Transmitting|Updating|(^$)' || true > } create subversion repo $ WCROOT="`pwd`/svn-wc" $ svnadmin create svn-repo $ svn co "$SVNREPOURL" svn-wc Checked out revision 0. $ cd svn-wc $ mkdir src $ echo alpha > src/alpha $ svn add src A src A src/alpha $ mkdir externals $ echo other > externals/other $ svn add externals A externals A externals/other $ svn ci -qm 'Add alpha' $ svn up -q $ echo "externals -r1 $SVNREPOURL/externals" > extdef $ svn propset -F extdef svn:externals src property 'svn:externals' set on 'src' $ svn ci -qm 'Setting externals' $ cd .. create hg repo $ mkdir sub $ cd sub $ hg init t $ cd t first revision, no sub $ echo a > a $ hg ci -Am0 adding a add first svn sub with leading whitespaces $ echo "s = [svn] $SVNREPOURL/src" >> .hgsub $ echo "subdir/s = [svn] $SVNREPOURL/src" >> .hgsub $ svn co --quiet "$SVNREPOURL"/src s $ mkdir subdir $ svn co --quiet "$SVNREPOURL"/src subdir/s $ hg add .hgsub svn subrepo is disabled by default $ hg ci -m1 abort: svn subrepos not allowed (see 'hg help config.subrepos' for details) [255] so enable it $ cat >> $HGRCPATH <<EOF > [subrepos] > svn:allowed = true > EOF $ hg ci -m1 make sure we avoid empty commits (issue2445) $ hg sum parent: 1:* tip (glob) 1 branch: default commit: (clean) update: (current) phases: 2 draft $ hg ci -moops nothing changed [1] debugsub $ hg debugsub path s source file:/*/$TESTTMP/svn-repo/src (glob) revision 2 path subdir/s source file:/*/$TESTTMP/svn-repo/src (glob) revision 2 change file in svn and hg, commit $ echo a >> a $ echo alpha >> s/alpha $ hg sum parent: 1:* tip (glob) 1 branch: default commit: 1 modified, 1 subrepos update: (current) phases: 2 draft $ hg commit --subrepos -m 'Message!' | filter_svn_output committing subrepository s Sending*s/alpha (glob) Committed revision 3. Fetching external item into '*s/externals'* (glob) External at revision 1. At revision 3. $ hg debugsub path s source file:/*/$TESTTMP/svn-repo/src (glob) revision 3 path subdir/s source file:/*/$TESTTMP/svn-repo/src (glob) revision 2 missing svn file, commit should fail $ rm s/alpha $ hg commit --subrepos -m 'abort on missing file' committing subrepository s abort: cannot commit missing svn entries (in subrepository "s") [255] $ svn revert s/alpha > /dev/null add an unrelated revision in svn and update the subrepo to without bringing any changes. $ svn mkdir "$SVNREPOURL/unrelated" -qm 'create unrelated' $ svn up -q s $ hg sum parent: 2:* tip (glob) Message! branch: default commit: (clean) update: (current) phases: 3 draft $ echo a > s/a should be empty despite change to s/a $ hg st add a commit from svn $ cd "$WCROOT/src" $ svn up -q $ echo xyz >> alpha $ svn propset svn:mime-type 'text/xml' alpha property 'svn:mime-type' set on 'alpha' $ svn ci -qm 'amend a from svn' $ cd ../../sub/t this commit from hg will fail $ echo zzz >> s/alpha $ (hg ci --subrepos -m 'amend alpha from hg' 2>&1; echo "[$?]") | grep -vi 'out of date' committing subrepository s abort: svn:*Commit failed (details follow): (glob) [255] $ svn revert -q s/alpha this commit fails because of meta changes $ svn propset svn:mime-type 'text/html' s/alpha property 'svn:mime-type' set on 's/alpha' $ (hg ci --subrepos -m 'amend alpha from hg' 2>&1; echo "[$?]") | grep -vi 'out of date' committing subrepository s abort: svn:*Commit failed (details follow): (glob) [255] $ svn revert -q s/alpha this commit fails because of externals changes $ echo zzz > s/externals/other $ hg ci --subrepos -m 'amend externals from hg' committing subrepository s abort: cannot commit svn externals (in subrepository "s") [255] $ hg diff --subrepos -r 1:2 | grep -v diff --- a/.hgsubstate Thu Jan 01 00:00:00 1970 +0000 +++ b/.hgsubstate Thu Jan 01 00:00:00 1970 +0000 @@ -1,2 +1,2 @@ -2 s +3 s 2 subdir/s --- a/a Thu Jan 01 00:00:00 1970 +0000 +++ b/a Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,2 @@ a +a $ svn revert -q s/externals/other this commit fails because of externals meta changes $ svn propset svn:mime-type 'text/html' s/externals/other property 'svn:mime-type' set on 's/externals/other' $ hg ci --subrepos -m 'amend externals from hg' committing subrepository s abort: cannot commit svn externals (in subrepository "s") [255] $ svn revert -q s/externals/other clone $ cd .. $ hg clone t tc updating to branch default A tc/s/alpha U tc/s Fetching external item into 'tc/s/externals'* (glob) A tc/s/externals/other Checked out external at revision 1. Checked out revision 3. A tc/subdir/s/alpha U tc/subdir/s Fetching external item into 'tc/subdir/s/externals'* (glob) A tc/subdir/s/externals/other Checked out external at revision 1. Checked out revision 2. 3 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cd tc debugsub in clone $ hg debugsub path s source file:/*/$TESTTMP/svn-repo/src (glob) revision 3 path subdir/s source file:/*/$TESTTMP/svn-repo/src (glob) revision 2 verify subrepo is contained within the repo directory $ "$PYTHON" -c "from __future__ import print_function; import os.path; print(os.path.exists('s'))" True update to nullrev (must delete the subrepo) $ hg up null 0 files updated, 0 files merged, 3 files removed, 0 files unresolved $ ls Check hg update --clean $ cd "$TESTTMP/sub/t" $ cd s $ echo c0 > alpha $ echo c1 > f1 $ echo c1 > f2 $ svn add f1 -q $ svn status | sort ? * a (glob) ? * f2 (glob) A * f1 (glob) M * alpha (glob) Performing status on external item at 'externals'* (glob) X * externals (glob) $ cd ../.. $ hg -R t update -C Fetching external item into 't/s/externals'* (glob) Checked out external at revision 1. Checked out revision 3. 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cd t/s $ svn status | sort ? * a (glob) ? * f1 (glob) ? * f2 (glob) Performing status on external item at 'externals'* (glob) X * externals (glob) Sticky subrepositories, no changes $ cd "$TESTTMP/sub/t" $ hg id -n 2 $ cd s $ svnversion 3 $ cd .. $ hg update 1 U *s/alpha (glob) Fetching external item into '*s/externals'* (glob) Checked out external at revision 1. Checked out revision 2. 2 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -n 1 $ cd s $ svnversion 2 $ cd .. Sticky subrepositories, file changes $ touch s/f1 $ cd s $ svn add f1 A f1 $ cd .. $ hg id -n 1+ $ cd s $ svnversion 2M $ cd .. $ hg update tip subrepository s diverged (local revision: 2, remote revision: 3) (M)erge, keep (l)ocal [working copy] or keep (r)emote [destination]? m subrepository sources for s differ use (l)ocal source (2) or (r)emote source (3)? l 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -n 2+ $ cd s $ svnversion 2M $ cd .. $ hg update --clean tip U *s/alpha (glob) Fetching external item into '*s/externals'* (glob) Checked out external at revision 1. Checked out revision 3. 1 files updated, 0 files merged, 0 files removed, 0 files unresolved Sticky subrepository, revision updates $ hg id -n 2 $ cd s $ svnversion 3 $ cd .. $ cd s $ svn update -qr 1 $ cd .. $ hg update 1 subrepository s diverged (local revision: 3, remote revision: 2) (M)erge, keep (l)ocal [working copy] or keep (r)emote [destination]? m subrepository sources for s differ (in checked out version) use (l)ocal source (1) or (r)emote source (2)? l 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -n 1+ $ cd s $ svnversion 1 $ cd .. Sticky subrepository, file changes and revision updates $ touch s/f1 $ cd s $ svn add f1 A f1 $ svnversion 1M $ cd .. $ hg id -n 1+ $ hg update tip subrepository s diverged (local revision: 3, remote revision: 3) (M)erge, keep (l)ocal [working copy] or keep (r)emote [destination]? m subrepository sources for s differ use (l)ocal source (1) or (r)emote source (3)? l 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -n 2+ $ cd s $ svnversion 1M $ cd .. Sticky repository, update --clean $ hg update --clean tip | grep -v 's[/\]externals[/\]other' U *s/alpha (glob) U *s (glob) Fetching external item into '*s/externals'* (glob) Checked out external at revision 1. Checked out revision 3. 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -n 2 $ cd s $ svnversion 3 $ cd .. Test subrepo already at intended revision: $ cd s $ svn update -qr 2 $ cd .. $ hg update 1 subrepository s diverged (local revision: 3, remote revision: 2) (M)erge, keep (l)ocal [working copy] or keep (r)emote [destination]? m 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg id -n 1+ $ cd s $ svnversion 2 $ cd .. Test case where subversion would fail to update the subrepo because there are unknown directories being replaced by tracked ones (happens with rebase). $ cd "$WCROOT/src" $ mkdir dir $ echo epsilon.py > dir/epsilon.py $ svn add dir A dir A dir/epsilon.py $ svn ci -qm 'Add dir/epsilon.py' $ cd ../.. $ hg init rebaserepo $ cd rebaserepo $ svn co -r5 --quiet "$SVNREPOURL"/src s $ echo "s = [svn] $SVNREPOURL/src" >> .hgsub $ hg add .hgsub $ hg ci -m addsub $ echo a > a $ hg add . adding a $ hg ci -m adda $ hg up 0 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ svn up -qr6 s $ hg ci -m updatesub created new head $ echo pyc > s/dir/epsilon.pyc $ hg up 1 D *s/dir (glob) Fetching external item into '*s/externals'* (glob) Checked out external at revision 1. Checked out revision 5. 2 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg up -q 2 Modify one of the externals to point to a different path so we can test having obstructions when switching branches on checkout: $ hg checkout tip 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ echo "obstruct = [svn] $SVNREPOURL/externals" >> .hgsub $ svn co -r5 --quiet "$SVNREPOURL"/externals obstruct $ hg commit -m 'Start making obstructed working copy' $ hg book other $ hg co -r 'p1(tip)' 2 files updated, 0 files merged, 0 files removed, 0 files unresolved (leaving bookmark other) $ echo "obstruct = [svn] $SVNREPOURL/src" >> .hgsub $ svn co -r5 --quiet "$SVNREPOURL"/src obstruct $ hg commit -m 'Other branch which will be obstructed' created new head Switching back to the head where we have another path mapped to the same subrepo should work if the subrepo is clean. $ hg co other A *obstruct/other (glob) Checked out revision 1. 2 files updated, 0 files merged, 0 files removed, 0 files unresolved (activating bookmark other) This is surprising, but is also correct based on the current code: $ echo "updating should (maybe) fail" > obstruct/other $ hg co tip abort: uncommitted changes (commit or update --clean to discard changes) [255] Point to a Subversion branch which has since been deleted and recreated First, create that condition in the repository. $ hg ci --subrepos -m cleanup | filter_svn_output committing subrepository obstruct Sending obstruct/other Committed revision 7. At revision 7. $ svn mkdir -qm "baseline" $SVNREPOURL/trunk $ svn copy -qm "initial branch" $SVNREPOURL/trunk $SVNREPOURL/branch $ svn co --quiet "$SVNREPOURL"/branch tempwc $ cd tempwc $ echo "something old" > somethingold $ svn add somethingold A somethingold $ svn ci -qm 'Something old' $ svn rm -qm "remove branch" $SVNREPOURL/branch $ svn copy -qm "recreate branch" $SVNREPOURL/trunk $SVNREPOURL/branch $ svn up -q $ echo "something new" > somethingnew $ svn add somethingnew A somethingnew $ svn ci -qm 'Something new' $ cd .. $ rm -rf tempwc $ svn co "$SVNREPOURL/branch"@10 recreated A recreated/somethingold Checked out revision 10. $ echo "recreated = [svn] $SVNREPOURL/branch" >> .hgsub $ hg ci -m addsub $ cd recreated $ svn up -q $ cd .. $ hg ci -m updatesub $ hg up -r-2 D *recreated/somethingnew (glob) A *recreated/somethingold (glob) Checked out revision 10. 1 files updated, 0 files merged, 0 files removed, 0 files unresolved (leaving bookmark other) $ test -f recreated/somethingold Test archive $ hg archive -S ../archive-all --debug --config progress.debug=true archiving: 0/2 files (0.00%) archiving: .hgsub 1/2 files (50.00%) archiving: .hgsubstate 2/2 files (100.00%) archiving (obstruct): 0/1 files (0.00%) archiving (obstruct): 1/1 files (100.00%) archiving (recreated): 0/1 files (0.00%) archiving (recreated): 1/1 files (100.00%) archiving (s): 0/2 files (0.00%) archiving (s): 1/2 files (50.00%) archiving (s): 2/2 files (100.00%) $ hg archive -S ../archive-exclude --debug --config progress.debug=true -X **old archiving: 0/2 files (0.00%) archiving: .hgsub 1/2 files (50.00%) archiving: .hgsubstate 2/2 files (100.00%) archiving (obstruct): 0/1 files (0.00%) archiving (obstruct): 1/1 files (100.00%) archiving (recreated): 0 files archiving (s): 0/2 files (0.00%) archiving (s): 1/2 files (50.00%) archiving (s): 2/2 files (100.00%) $ find ../archive-exclude | sort ../archive-exclude ../archive-exclude/.hg_archival.txt ../archive-exclude/.hgsub ../archive-exclude/.hgsubstate ../archive-exclude/obstruct ../archive-exclude/obstruct/other ../archive-exclude/s ../archive-exclude/s/alpha ../archive-exclude/s/dir ../archive-exclude/s/dir/epsilon.py Test forgetting files, not implemented in svn subrepo, used to traceback $ hg forget 'notafile*' notafile*: $ENOENT$ [1] Test a subrepo referencing a just moved svn path. Last commit rev will be different from the revision, and the path will be different as well. $ cd "$WCROOT" $ svn up > /dev/null $ mkdir trunk/subdir branches $ echo a > trunk/subdir/a $ svn add trunk/subdir branches A trunk/subdir A trunk/subdir/a A branches $ svn ci -qm addsubdir $ svn cp -qm branchtrunk $SVNREPOURL/trunk $SVNREPOURL/branches/somebranch $ cd .. $ hg init repo2 $ cd repo2 $ svn co $SVNREPOURL/branches/somebranch/subdir A subdir/a Checked out revision 15. $ echo "subdir = [svn] $SVNREPOURL/branches/somebranch/subdir" > .hgsub $ hg add .hgsub $ hg ci -m addsub $ hg up null 0 files updated, 0 files merged, 2 files removed, 0 files unresolved $ hg up A *subdir/a (glob) Checked out revision 15. 2 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cd .. Test sanitizing ".hg/hgrc" in subrepo $ cd sub/t $ hg update -q -C tip $ cd s $ mkdir .hg $ echo '.hg/hgrc in svn repo' > .hg/hgrc $ mkdir -p sub/.hg $ echo 'sub/.hg/hgrc in svn repo' > sub/.hg/hgrc $ svn add .hg sub A .hg A .hg/hgrc A sub A sub/.hg A sub/.hg/hgrc $ svn ci -qm 'add .hg/hgrc to be sanitized at hg update' $ svn up -q $ cd .. $ hg commit -S -m 'commit with svn revision including .hg/hgrc' $ grep ' s$' .hgsubstate 16 s $ cd .. $ hg -R tc pull -u -q 2>&1 | sort warning: removing potentially hostile 'hgrc' in '$TESTTMP/sub/tc/s/.hg' warning: removing potentially hostile 'hgrc' in '$TESTTMP/sub/tc/s/sub/.hg' $ cd tc $ grep ' s$' .hgsubstate 16 s $ test -f s/.hg/hgrc [1] $ test -f s/sub/.hg/hgrc [1] Test that sanitizing is omitted in meta data area: $ mkdir s/.svn/.hg $ echo '.hg/hgrc in svn metadata area' > s/.svn/.hg/hgrc $ hg update -q -C '.^1' $ cd ../.. SEC: test for ssh exploit $ hg init ssh-vuln $ cd ssh-vuln $ echo "s = [svn]$SVNREPOURL/src" >> .hgsub $ svn co --quiet "$SVNREPOURL"/src s $ hg add .hgsub $ hg ci -m1 $ echo "s = [svn]svn+ssh://-oProxyCommand=touch%20owned%20nested" > .hgsub $ hg ci -m2 $ cd .. $ hg clone ssh-vuln ssh-vuln-clone updating to branch default abort: potentially unsafe url: 'svn+ssh://-oProxyCommand=touch owned nested' (in subrepository "s") [255] also check that a percent encoded '-' (%2D) doesn't work $ cd ssh-vuln $ echo "s = [svn]svn+ssh://%2DoProxyCommand=touch%20owned%20nested" > .hgsub $ hg ci -m3 $ cd .. $ rm -r ssh-vuln-clone $ hg clone ssh-vuln ssh-vuln-clone updating to branch default abort: potentially unsafe url: 'svn+ssh://-oProxyCommand=touch owned nested' (in subrepository "s") [255] also check that hiding the attack in the username doesn't work: $ cd ssh-vuln $ echo "s = [svn]svn+ssh://%2DoProxyCommand=touch%20owned%20foo@example.com/nested" > .hgsub $ hg ci -m3 $ cd .. $ rm -r ssh-vuln-clone $ hg clone ssh-vuln ssh-vuln-clone updating to branch default abort: potentially unsafe url: 'svn+ssh://-oProxyCommand=touch owned foo@example.com/nested' (in subrepository "s") [255]