Page MenuHomePhabricator

joerg.sonnenberger (Joerg Sonnenberger)
User

Projects

User does not belong to any projects.

User Details

User Since
Dec 6 2017, 9:25 AM (232 w, 21 h)

Recent Activity

Tue, May 17

joerg.sonnenberger added a comment to D12626: revlog: use %d to format int instead of %lu (issue6565).

I think there are two issues here. The format string in line 500 doesn't match 492 and the format string in 494 doesn't either.

Tue, May 17, 2:48 PM

Thu, Apr 21

joerg.sonnenberger added a comment to D12583: iblt: prototype for setdiscovery.

This is a proof-of-concept still, not for actual merge.

Thu, Apr 21, 7:27 PM
joerg.sonnenberger created D12583: iblt: prototype for setdiscovery.
Thu, Apr 21, 7:25 PM

Apr 6 2022

joerg.sonnenberger committed rHG5b65721a75eb: revlog: recommit 49fd21f32695 with a fix for issue6528.
revlog: recommit 49fd21f32695 with a fix for issue6528
Apr 6 2022, 6:09 AM
joerg.sonnenberger closed D11203: revlog: recommit 49fd21f32695 with a fix for issue6528.
Apr 6 2022, 5:46 AM
joerg.sonnenberger committed rHG1b95e967e771: revlog: recommit 49fd21f32695 with a fix for issue6528.
revlog: recommit 49fd21f32695 with a fix for issue6528
Apr 6 2022, 5:45 AM

Mar 29 2022

joerg.sonnenberger added a comment to D11203: revlog: recommit 49fd21f32695 with a fix for issue6528.

default should be fine. There are a number of places that don't necessarily work correctly for miss-ordered parents, but that's been the default behavior for years too, so nothing new.

Mar 29 2022, 8:57 AM
joerg.sonnenberger updated the diff for D11203: revlog: recommit 49fd21f32695 with a fix for issue6528.
Mar 29 2022, 8:56 AM
joerg.sonnenberger updated the diff for D11203: revlog: recommit 49fd21f32695 with a fix for issue6528.
Mar 29 2022, 8:48 AM
joerg.sonnenberger updated the diff for D12391: pullbundle: fix file name in the help text.
Mar 29 2022, 8:27 AM

Mar 27 2022

joerg.sonnenberger added a comment to D12413: stringutil: try to avoid running `splitlines()` only to get first line.

Oh, sorry. I thought you returned the slice directly, which should avoid some extra memory allocations.

Mar 27 2022, 4:18 PM
joerg.sonnenberger added a comment to D12413: stringutil: try to avoid running `splitlines()` only to get first line.

Well, for DOS-style line ending, the find will point to the \n and we should return everything before the \r. If we ignore line endings mixed with old Mac-style, just checking the character before is enough to cover both DOS and Unix convention.

Mar 27 2022, 3:34 PM
joerg.sonnenberger added a comment to D12413: stringutil: try to avoid running `splitlines()` only to get first line.

Note that you should at least try to check if text[i-1] is \r, otherwise the result changes from before. I'm perfectly fine with not supporting the ancient Mac convention of using \r only, but we should work properly for DOS-style \r\n.

Mar 27 2022, 3:08 PM

Mar 25 2022

joerg.sonnenberger added a comment to D12404: templates: extract function to `stringutil` for getting first line of text.

I wonder if we shouldn't optimize for the common case of multiple lines by using find() instead. Searching for \n first and \r second (in the substring if the first got a hit) and only then slicing the result.

Mar 25 2022, 7:09 AM

Mar 24 2022

joerg.sonnenberger added inline comments to D12402: streamclone: avoid some obscure error in a corner case.
Mar 24 2022, 8:30 PM

Mar 22 2022

joerg.sonnenberger closed D12391: pullbundle: fix file name in the help text.
Mar 22 2022, 9:37 AM
joerg.sonnenberger committed rHG6e7ae547e38d: pullbundle: fix file name in the help text.
pullbundle: fix file name in the help text
Mar 22 2022, 9:35 AM
joerg.sonnenberger added a comment to D12394: subsetmaker: stabilize the computation of `scratch` subset.

Why not pickable = sorted(heads)?

Mar 22 2022, 9:15 AM

Mar 21 2022

joerg.sonnenberger created D12391: pullbundle: fix file name in the help text.
Mar 21 2022, 10:21 PM

Mar 2 2022

joerg.sonnenberger added a comment to D12288: posix: delete Python 2 posixfile().

Please fix the comment to something like: Python 3 file objects set the initial position consistently for append mode and can be used directly or something like that. The current comment doesn't make sense without the context.

Mar 2 2022, 6:47 PM

Feb 21 2022

joerg.sonnenberger created D12207: zstd: hack include order to ensure that our zstd.h is found.
Feb 21 2022, 8:45 AM

Feb 7 2022

joerg.sonnenberger added a comment to D12139: rank: naive rank property computation and retrieval.

Why do we want this variant of a rank? E.g. an alternative definition of rank is the length of the longest path to root. That definition is much easier to compute incrementally and just as useful for most algorithmic uses.

Feb 7 2022, 3:01 PM
joerg.sonnenberger added a comment to D12137: revlog: split revlog v1 and revlog v2 handling.

Most of v1 and v2 is shared and duplicating all of that just adds complications for no gain.

Feb 7 2022, 9:26 AM
joerg.sonnenberger added a comment to D12136: revlog: extract entry byte offsets into named constants.

This is reasonable, but I didn't double-check that all cases are identical.

Feb 7 2022, 7:00 AM
joerg.sonnenberger added a comment to D12137: revlog: split revlog v1 and revlog v2 handling.

I'm not a fan of this change, it seems a step in the wrong direction.

Feb 7 2022, 6:59 AM

Jan 27 2022

joerg.sonnenberger added a comment to D12096: stream-clone: allow to change persistent-nodemap format during stream clone.

"fast to general locally" in the description needs a fix. I wonder if we should introduce a new set for such features, I expect similar cases to occur in the future.

Jan 27 2022, 7:34 PM

Jan 17 2022

joerg.sonnenberger closed D11956: exchange: add fast path for subrepo check on push.
Jan 17 2022, 5:54 AM
joerg.sonnenberger committed rHG28f0092ec89f: exchange: add fast path for subrepo check on push.
exchange: add fast path for subrepo check on push
Jan 17 2022, 5:53 AM

Jan 8 2022

joerg.sonnenberger updated the diff for D11956: exchange: add fast path for subrepo check on push.
Jan 8 2022, 5:33 AM

Jan 3 2022

joerg.sonnenberger added inline comments to D11956: exchange: add fast path for subrepo check on push.
Jan 3 2022, 7:01 AM

Jan 2 2022

joerg.sonnenberger created D11956: exchange: add fast path for subrepo check on push.
Jan 2 2022, 7:30 PM

Dec 23 2021

joerg.sonnenberger accepted D11809: docs: add missing dirstate-v2 documentation.
Dec 23 2021, 10:11 AM
joerg.sonnenberger accepted D11811: docs: update Rust readme with a mention of `rhg`.
Dec 23 2021, 9:11 AM
joerg.sonnenberger accepted D11810: docs: fix typos and wording for the persistent-nodemap feature.
Dec 23 2021, 9:10 AM
joerg.sonnenberger added inline comments to D11809: docs: add missing dirstate-v2 documentation.
Dec 23 2021, 9:09 AM
joerg.sonnenberger added inline comments to D11808: docs: add documentation about Rust.
Dec 23 2021, 9:06 AM

Sep 18 2021

joerg.sonnenberger created D11453: log: if there is a merge state, show conflictother for . [WIP].
Sep 18 2021, 7:51 AM

Sep 17 2021

joerg.sonnenberger added a comment to D7177: rebase: introduce optional parent mapping.

From the VC call today: The use case can be addressed by marking parents as to-be-preserved, so that for merge commits rebase can decide on the parent link is should operate on. It covers a general DAG and is somewhat easier to use in terms of UX than the parentmap, but can provide the same functionality with a multi-step rebase.

Sep 17 2021, 9:59 AM

Aug 4 2021

joerg.sonnenberger created D11252: debugupgraderepo: add fix-metaencoding-flag pass for issue6528.
Aug 4 2021, 4:42 PM

Jul 20 2021

joerg.sonnenberger added a comment to D11203: revlog: recommit 49fd21f32695 with a fix for issue6528.

I would have to reaudit the parents use, but it wasn't just new code that misbehaves on misordered parents. E.g. memctx.parents for a trivial example. So yes, just backing it out introduces regressions as well.

Jul 20 2021, 3:36 PM
joerg.sonnenberger created D11203: revlog: recommit 49fd21f32695 with a fix for issue6528.
Jul 20 2021, 9:58 AM

Jul 6 2021

joerg.sonnenberger added a comment to D10995: corruption: backout changeset 49fd21f32695 (issue6528).

Just the revert as is introduces other regressions. At the very least, we should preserve the sorting for changelog and keep the original order for filelog. The other part is that the filelog logic here is actively broken even before, so this doesn't fix the logic, just makes it less visibly broken.

Jul 6 2021, 3:09 PM

Jun 21 2021

joerg.sonnenberger added a comment to D10888: copies: Keep changelog sidedata file open during copy tracing.

Can you comment on how many of them we keep open at a time?

Jun 21 2021, 5:49 AM

May 27 2021

joerg.sonnenberger added a comment to D10538: tests: allow Google's internal builds of clang-format to be used.

Consider it more a -0, not a very strong objection.

May 27 2021, 6:27 PM

May 20 2021

joerg.sonnenberger added a comment to D10712: tests: show that hg fails to rollback transaction on revlog split.

Correct, this stack should have beeen superseded by those.

May 20 2021, 9:23 AM
joerg.sonnenberger committed rHG46b828b85eb7: revlog: update data file record before index rename.
revlog: update data file record before index rename
May 20 2021, 4:36 AM
joerg.sonnenberger committed rHG18415fc918a1: recover: only apply last journal record per file (issue6423).
recover: only apply last journal record per file (issue6423)
May 20 2021, 4:36 AM
joerg.sonnenberger committed rHG21ed126bab53: revlog: fix index computation during inline->non-inline transition.
revlog: fix index computation during inline->non-inline transition
May 20 2021, 4:36 AM
joerg.sonnenberger committed rHG25f42db82ac8: revlog: update data file record before index rename.
revlog: update data file record before index rename
May 20 2021, 3:52 AM
joerg.sonnenberger committed rHGbca9faa0aafc: recover: only apply last journal record per file (issue6423).
recover: only apply last journal record per file (issue6423)
May 20 2021, 3:52 AM
joerg.sonnenberger committed rHGf4117223239b: revlog: fix index computation during inline->non-inline transition.
revlog: fix index computation during inline->non-inline transition
May 20 2021, 3:51 AM

May 19 2021

joerg.sonnenberger closed D10725: revlog: update data file record before index rename.
May 19 2021, 1:58 PM
joerg.sonnenberger closed D10724: revlog: fix index computation during inline->non-inline transition.
May 19 2021, 1:58 PM
joerg.sonnenberger closed D10726: recover: only apply last journal record per file (issue6423).
May 19 2021, 1:58 PM
joerg.sonnenberger committed rHG672f48cf1d03: recover: only apply last journal record per file (issue6423).
recover: only apply last journal record per file (issue6423)
May 19 2021, 1:58 PM
joerg.sonnenberger committed rHGd92310d45ecb: revlog: update data file record before index rename.
revlog: update data file record before index rename
May 19 2021, 1:58 PM
joerg.sonnenberger committed rHG762afde8c3a0: revlog: fix index computation during inline->non-inline transition.
revlog: fix index computation during inline->non-inline transition
May 19 2021, 1:57 PM
joerg.sonnenberger retitled D10726: recover: only apply last journal record per file (issue6423) from recover: only apply last journal record per file to recover: only apply last journal record per file (issue6423).
May 19 2021, 9:54 AM
joerg.sonnenberger updated the diff for D10726: recover: only apply last journal record per file (issue6423).
May 19 2021, 9:51 AM
joerg.sonnenberger updated the diff for D10725: revlog: update data file record before index rename.
May 19 2021, 7:48 AM

May 18 2021

joerg.sonnenberger updated the summary of D10725: revlog: update data file record before index rename.
May 18 2021, 2:48 PM
joerg.sonnenberger updated the summary of D10724: revlog: fix index computation during inline->non-inline transition.
May 18 2021, 2:48 PM
joerg.sonnenberger updated the diff for D10726: recover: only apply last journal record per file (issue6423).
May 18 2021, 2:48 PM

May 17 2021

joerg.sonnenberger created D10726: recover: only apply last journal record per file (issue6423).
May 17 2021, 10:51 PM
joerg.sonnenberger created D10725: revlog: update data file record before index rename.
May 17 2021, 10:27 PM
joerg.sonnenberger updated the diff for D10724: revlog: fix index computation during inline->non-inline transition.
May 17 2021, 9:09 PM
joerg.sonnenberger created D10724: revlog: fix index computation during inline->non-inline transition.
May 17 2021, 8:42 PM

May 16 2021

joerg.sonnenberger added a comment to D10712: tests: show that hg fails to rollback transaction on revlog split.

There are three stages for the revlog and I think we need to teach the transaction playback code at least a bit about this to handle it correctly:

May 16 2021, 9:55 PM
D10712: tests: show that hg fails to rollback transaction on revlog split now requires changes to proceed.

This whole stack exposes a real problem and one quite a bit older, but it is not correct in a number of ways as is.

May 16 2021, 9:03 PM

May 15 2021

joerg.sonnenberger accepted D10711: hgweb: Fix deprecation warning in Python 3.10 (issue6520).
May 15 2021, 5:41 PM

May 14 2021

joerg.sonnenberger added a comment to D10710: vfs: Fix deprecation warning in Python 3.10 (issue6520).

Silly Python, but this should be fine for all versions we care about. Ironically, it used to be defined the other way around as alias...

May 14 2021, 7:14 AM

May 11 2021

joerg.sonnenberger accepted D10707: templates: fix `revset('parents()') % ...` in amend message template.
May 11 2021, 3:55 PM
D10707: templates: fix `revset('parents()') % ...` in amend message template now requires changes to proceed.
May 11 2021, 3:41 PM

May 4 2021

joerg.sonnenberger added a comment to D10624: revlogv2: introduce a very basic docket file.

Adding more files to the revlog layer sounds like a move in the wrong direction and the motivation here seems quite weak too. Please discuss this with an actual plan on the mailing list.

May 4 2021, 1:06 PM
joerg.sonnenberger added a comment to D10646: revlog: add a "data compression mode" entry in the index tuple.

I don't understand the motivation for this change.

May 4 2021, 1:03 PM
joerg.sonnenberger added a comment to D10650: revlog: introduce a plain compression mode.

I don't understand this change. We already support none as compression method.

May 4 2021, 1:02 PM

May 3 2021

joerg.sonnenberger closed D10535: core: don't hard-code hex node lengths.
May 3 2021, 9:36 AM
joerg.sonnenberger closed D10536: core: don't hard-code node length.
May 3 2021, 9:35 AM
joerg.sonnenberger closed D10533: manifests: push down expected node length into the parser.
May 3 2021, 9:35 AM
joerg.sonnenberger committed rHG12450fbea288: manifests: push down expected node length into the parser.
manifests: push down expected node length into the parser
May 3 2021, 9:34 AM
joerg.sonnenberger committed rHGc5e1cc0b4c77: core: don't hard-code node length.
core: don't hard-code node length
May 3 2021, 9:34 AM
joerg.sonnenberger committed rHGa407fe56d6e8: core: don't hard-code hex node lengths.
core: don't hard-code hex node lengths
May 3 2021, 9:33 AM
joerg.sonnenberger added a comment to D10593: revlog: simplify a conditionnal in _enforceinlinesize.

This adds a penalty for the non-inline case?

May 3 2021, 9:20 AM
joerg.sonnenberger closed D10534: tests: bump default timeout to 360s.
May 3 2021, 8:49 AM
joerg.sonnenberger committed rHGc853bc60c5d6: tests: bump default timeout to 360s.
tests: bump default timeout to 360s
May 3 2021, 8:49 AM

Apr 30 2021

joerg.sonnenberger added a comment to D10538: tests: allow Google's internal builds of clang-format to be used.

I'm not sure that this is an improvement. We discussed whether we want to hard-wire the version even stricter ("exactly version 11"), especially because it is hard to predict when the output is going to change again.

Apr 30 2021, 8:17 PM

Apr 29 2021

joerg.sonnenberger created D10536: core: don't hard-code node length.
Apr 29 2021, 9:21 PM
joerg.sonnenberger created D10535: core: don't hard-code hex node lengths.
Apr 29 2021, 9:11 PM
joerg.sonnenberger created D10534: tests: bump default timeout to 360s.
Apr 29 2021, 8:26 PM
joerg.sonnenberger created D10533: manifests: push down expected node length into the parser.
Apr 29 2021, 8:14 PM
joerg.sonnenberger added a comment to D10529: node: replace nullid and friends with nodeconstants class.

Already merged as d55b71393907

Apr 29 2021, 9:29 AM

Apr 28 2021

joerg.sonnenberger committed rHG77e73827a02d: tests: don't hard-code /bin/bash.
tests: don't hard-code /bin/bash
Apr 28 2021, 11:09 AM
joerg.sonnenberger committed rHG8d2b62d716b0: packaging: extract pre-computed version when running from plain tarball.
packaging: extract pre-computed version when running from plain tarball
Apr 28 2021, 11:08 AM
joerg.sonnenberger committed rHGd55b71393907: node: replace nullid and friends with nodeconstants class.
node: replace nullid and friends with nodeconstants class
Apr 28 2021, 11:08 AM
joerg.sonnenberger added a comment to D10502: packaging: extract pre-computed version when running from plain tarball.

Merged as 8d2b62d716b0

Apr 28 2021, 11:07 AM
joerg.sonnenberger added a comment to D10507: tests: don't hard-code /bin/bash.

Merged as 77e73827a02d

Apr 28 2021, 11:07 AM
joerg.sonnenberger created D10529: node: replace nullid and friends with nodeconstants class.
Apr 28 2021, 11:07 AM
joerg.sonnenberger updated the diff for D9465: node: replace nullid and friends with nodeconstants class.
Apr 28 2021, 11:06 AM
joerg.sonnenberger created D10507: tests: don't hard-code /bin/bash.
Apr 28 2021, 10:59 AM

Apr 20 2021

joerg.sonnenberger added inline comments to D10502: packaging: extract pre-computed version when running from plain tarball.
Apr 20 2021, 8:09 PM