s.
I've attached a patch.
[1] https://www.sqlite.org/lang_conflict.html
Regards,
Denis Kovalchuk
rep-cache.db insert optimization.
Use the 'IGNORE' conflict resolution algorithm [1] for STMT_SET_REP
and remove SVN_ERR_SQLITE_CONSTRAINT handling, because the error
should not occur now. It brings a slig
as?
I started to handle this problem and plan to report the result.
Regards,
Denis Kovalchuk
ould be skipped
if read_rep_cache() is not supported by Python-SQLite version.
I have attached a patch that fixes this.
Regards,
Denis Kovalchuk
Follow-up to r1875921: Add missing test decorator.
Skip build_repcache() test if read_rep_cache() is not supported
by Python-SQLite version.
* subversio
st commit
processing stage.
Regards,
Denis Kovalchuk
of using the ‘IGNORE’ algorithm: ~15000 microseconds.
Outside of the benchmark, there is ~15% improvement for the
‘build-repcache’ command.
Based on this, I assume that there are other commands that may benefit from
this change.
I have attached an updated patch.
Regards,
Denis Kovalchuk
> > This seems to fix it.
> >
> > I don't understand why this is necessary but the regular code path for
> > access to node revision data in fsfs also seems to apply this always.
> >
> > Can an fsfs expert confirm?
>
> Since this fix looks reasonable to me and we're trying to move the release
>
attached a patch that fixes it.
Regards,
Denis Kovalchuk
Fix undefined behavior when constructing ID for txn_node_cache in fsfs.
In make_txn_root() function, the txn variable of type svn_fs_fs__id_part_t *
is passed to apr_pstrcat() function, leading to the undefined behavior.
I can assume
.
Regards,
Denis Kovalchuk
Fix "database is locked" error when committing during rep-cache.db
verification.
If a commit occurs during the rep-cache.db verification, this can lead to a
post-commit error "database is locked" and new representations will not be
added in the re
tion, the
new approach has a guarantee that all entries that existed at the time of the
call of 'svnadmin verify' will be verified.
So I think that it should not affect the correctness of verification.
Regards,
Denis Kovalchuk
ds only on the values in that row. These
values are immutable. The revision content on fsfs is also immutable. Based on
this, I assume that if we do not find errors in accessed rows, there are no
errors in rows that existed at the time of the call of 'svnadmin verify'.
Regards,
Denis Kovalchuk
s an improvement of the current one, which
also solves the problem. But if there are still significant concerns about that
approach, I do not have anything else to offer at the moment. I may try to
return to this topic if I have any additional thoughts.
Regards,
Denis Kovalchuk
possible to provide a
similar behavior walking the array of iprop requests from the end and get rid
of extra copying of array elements.
I've attached a patch.
Regards,
Denis Kovalchuk
Fix an inefficient way to fill an array of inherited properties.
The code that obtains inherited properties
t which doesn't require running «build-repcache» afterwards;
Speaking about tradeoffs, I would like to note that these cases are not
equivalent, when it comes to visible behavior, because the requirement to run
build-repcache does not fix 1), 3) and 4).
[1] https://www.sqlite.org/lang_attach.html
[2
with the implementation of the command.
[1] https://svn.haxx.se/users/archive-2014-01/0128.shtml
[2] https://svn.haxx.se/dev/archive-2018-08/0050.shtml
Regards,
Denis Kovalchuk
Introduce 'svnadmin build-repcache' command.
Implement the 'svnadmin build-repcache' CLI and add an ioctl API for building
nnot have only
fs_has_rep_sharing()
call for the test, because fsx also supports rep-sharing, but we do not
have an implementation for it. I have attached an updated patch. Regards, Denis
Kovalchuk
On Wed, Mar 18, 2020 at 7:22 PM Daniel Shahaf
wrote:
> Denis Kovalchuk wrote on Wed, 18 Mar 20
test's decorators.
>
I think that this might be unrelated to this patch, but also I may be
misunderstanding
what exactly you propose. If we modify fs_has_rep_sharing() to return True
for FSFS and to return False for other backends, we will have a problem
with FSX,
which also supports rep-sharin
eisure and delete it when done with it?
I think that compared to the patch this approach has a problem that every
verify will lead to a lot of write operations. For example, the size of
rep-cache.db in the ASF repository is 1.1 GB.
Just in case, I will be happy to work on any specific things in the patch, if
there are some things you think can be improved.
Regards,
Denis Kovalchuk
yle":
svn info
...
Working Copy Root Path: C:\Projects\subversion
...
I have attached a patch that adds a missing svn_dirent_local_style() call to
the print_info_item() function, where the corresponding wc-root path is
processed.
Best Regards,
Denis Kovalchuk
Fix: svn info outputs wc-root path i
18 matches
Mail list logo
