Page MenuHomePhabricator

singhsrb (Saurabh Singh)
User

Projects

User does not belong to any projects.

User Details

User Since
Aug 31 2017, 8:20 PM (159 w, 1 d)

Recent Activity

Jan 31 2018

singhsrb abandoned D1817: treemanifest: separate out logic to write client tree manifest.

Will be rebasing and floating for review again.

Jan 31 2018, 5:45 PM
singhsrb abandoned D1818: treemanifest: use flat manifest hash during client tree manifest writes.

Will be rebasing and floating for review again.

Jan 31 2018, 5:45 PM
singhsrb abandoned D1819: treemanifest: write client tree manifests eagerly during nested transactions.

Will be rebasing and floating for review again.

Jan 31 2018, 5:45 PM
singhsrb abandoned D1708: treemanifest: store ui once instead of using repo.ui every time.

Will be rebasing and floating for review again.

Jan 31 2018, 5:45 PM
singhsrb abandoned D1709: remotetreedatastore: make MissingNodesError subclass of KeyError.

Will be rebasing and floating for review again.

Jan 31 2018, 5:45 PM
singhsrb abandoned D1822: test-treemanifest-treeonly: update test to have more flat manifests.

Will be rebasing and floating for review again.

Jan 31 2018, 5:44 PM
singhsrb abandoned D1823: backfill: refactor to segregate logic for converting manifests.

Will be rebasing and floating for review again.

Jan 31 2018, 5:44 PM
singhsrb abandoned D1824: treemanifest: refactor write manifest wrapper.

Will be rebasing and floating for review again.

Jan 31 2018, 5:44 PM
singhsrb abandoned D1825: treemanifest: add client support for flat manifest to tree manifest.

Will be rebasing and floating for review again.

Jan 31 2018, 5:44 PM
singhsrb abandoned D1826: treemanifest: demand convert flat manifest to tree manifest in treeonly mode.

Will be rebasing and floating for review again.

Jan 31 2018, 5:43 PM

Jan 8 2018

singhsrb updated the test plan for D1818: treemanifest: use flat manifest hash during client tree manifest writes.
Jan 8 2018, 9:54 AM
singhsrb updated the test plan for D1825: treemanifest: add client support for flat manifest to tree manifest.
Jan 8 2018, 9:52 AM
singhsrb updated the test plan for D1819: treemanifest: write client tree manifests eagerly during nested transactions.
Jan 8 2018, 9:50 AM

Jan 5 2018

singhsrb created D1826: treemanifest: demand convert flat manifest to tree manifest in treeonly mode.
Jan 5 2018, 11:50 PM
singhsrb updated the diff for D1823: backfill: refactor to segregate logic for converting manifests.
Jan 5 2018, 8:27 PM
singhsrb created D1825: treemanifest: add client support for flat manifest to tree manifest.
Jan 5 2018, 8:26 PM
singhsrb retitled D1824: treemanifest: refactor write manifest wrapper from treemanifest: refactor write manifest wrapper to avoid redundancy to treemanifest: refactor write manifest wrapper.
Jan 5 2018, 8:09 PM
singhsrb created D1824: treemanifest: refactor write manifest wrapper.
Jan 5 2018, 8:08 PM
singhsrb created D1823: backfill: refactor to segregate logic for converting manifests.
Jan 5 2018, 8:00 PM
singhsrb created D1822: test-treemanifest-treeonly: update test to have more flat manifests.
Jan 5 2018, 7:26 PM
singhsrb created D1821: treemanifest: subclass remotetreedatastore from generatingdatastore.
Jan 5 2018, 7:18 PM
singhsrb created D1820: treemanifest: refactor remotetreedatastore call hierarchy.
Jan 5 2018, 6:44 PM
singhsrb created D1819: treemanifest: write client tree manifests eagerly during nested transactions.
Jan 5 2018, 6:41 PM
singhsrb created D1818: treemanifest: use flat manifest hash during client tree manifest writes.
Jan 5 2018, 6:28 PM
singhsrb created D1817: treemanifest: separate out logic to write client tree manifest.
Jan 5 2018, 6:27 PM

Dec 16 2017

singhsrb added a comment to D1709: remotetreedatastore: make MissingNodesError subclass of KeyError.

the test test-treemanifest-infinitepush.t (included in this commit) reflects the change in behavior because of the change in error hierarchy. Am I missing something?

Dec 16 2017, 12:39 AM

Dec 15 2017

singhsrb updated the diff for D1709: remotetreedatastore: make MissingNodesError subclass of KeyError.
Dec 15 2017, 7:59 PM
singhsrb added inline comments to D1709: remotetreedatastore: make MissingNodesError subclass of KeyError.
Dec 15 2017, 7:52 PM
singhsrb updated the diff for D1708: treemanifest: store ui once instead of using repo.ui every time.
Dec 15 2017, 7:25 PM
singhsrb added inline comments to D1708: treemanifest: store ui once instead of using repo.ui every time.
Dec 15 2017, 7:17 PM
singhsrb created D1709: remotetreedatastore: make MissingNodesError subclass of KeyError.
Dec 15 2017, 4:11 PM
singhsrb created D1708: treemanifest: store ui once instead of using repo.ui every time.
Dec 15 2017, 3:33 PM
singhsrb closed D1707: tests: use testrepohg in test-check tests.
Dec 15 2017, 3:24 PM
singhsrb committed rFBHGX32ceeccb832c: tests: use testrepohg in test-check tests.
tests: use testrepohg in test-check tests
Dec 15 2017, 3:23 PM
singhsrb updated the diff for D1707: tests: use testrepohg in test-check tests.
Dec 15 2017, 3:17 PM
singhsrb accepted D1707: tests: use testrepohg in test-check tests.

Thanks for the change! I added a few more places where this is needed for all the tests to pass on my machine.

Dec 15 2017, 3:14 PM

Dec 12 2017

singhsrb closed D1596: incremental-repack: do not repack largest 2 files if nothing else to do.
Dec 12 2017, 7:25 PM
singhsrb committed rFBHGX6e975e37404d: prefetch: fix the hg executable location for background prefetch.
prefetch: fix the hg executable location for background prefetch
Dec 12 2017, 7:25 PM
singhsrb committed rFBHGX21fd2c088d1d: incremental-repack: do not repack largest 2 files if nothing else to do.
incremental-repack: do not repack largest 2 files if nothing else to do
Dec 12 2017, 7:25 PM
singhsrb closed D1600: prefetch: fix the hg executable location for background prefetch.
Dec 12 2017, 7:25 PM

Dec 7 2017

singhsrb added inline comments to D1614: noverify: add extension to skip repo verification.
Dec 7 2017, 9:51 PM
singhsrb added inline comments to D1614: noverify: add extension to skip repo verification.
Dec 7 2017, 8:46 PM

Dec 6 2017

singhsrb added a comment to D1484: incremental-repack: prefer small packs across generations over largest ones.

D1588 and D1596 address the major ideas from this commit.

Dec 6 2017, 1:48 PM

Dec 5 2017

singhsrb added a comment to D1600: prefetch: fix the hg executable location for background prefetch.

Basically the same issue as D1221. I am not even sure if this is used anywhere but I guess no harm in fixing it.

Dec 5 2017, 5:21 PM
singhsrb added a comment to D1221: repack: fix the hg executable location for background repack.
Dec 5 2017, 5:21 PM
singhsrb created D1600: prefetch: fix the hg executable location for background prefetch.
Dec 5 2017, 5:17 PM
singhsrb added a comment to D1596: incremental-repack: do not repack largest 2 files if nothing else to do.

Note that this will work particularly well with D1588 because we will move towards the max pack file size limit introduced there in bigger steps.

Dec 5 2017, 4:25 PM
singhsrb created D1596: incremental-repack: do not repack largest 2 files if nothing else to do.
Dec 5 2017, 4:19 PM
singhsrb closed D1588: incremental-repack: do not repack files over a configuration based size.
Dec 5 2017, 4:05 PM
singhsrb closed D1586: incremental-repack: adding documentation for some repacking options.
Dec 5 2017, 4:05 PM
singhsrb closed D1587: incremental-repack: refactor to facilitate easier options manipulation.
Dec 5 2017, 4:05 PM
singhsrb committed rFBHGX69ccc796d4ba: incremental-repack: do not repack files over a configuration based size.
incremental-repack: do not repack files over a configuration based size
Dec 5 2017, 4:05 PM
singhsrb committed rFBHGX14fa93b8f742: incremental-repack: refactor to facilitate easier options manipulation.
incremental-repack: refactor to facilitate easier options manipulation
Dec 5 2017, 4:05 PM
singhsrb committed rFBHGX0932b10d62d5: incremental-repack: adding documentation for some repacking options.
incremental-repack: adding documentation for some repacking options
Dec 5 2017, 4:05 PM
singhsrb added inline comments to D1588: incremental-repack: do not repack files over a configuration based size.
Dec 5 2017, 2:46 PM

Dec 4 2017

singhsrb closed D1569: basestore: remove unnecessary files with suffix '_old' during clean up.
Dec 4 2017, 9:00 PM
singhsrb closed D1571: basestore: use tryunlink from mercurial.util to try and remove files.
Dec 4 2017, 9:00 PM
singhsrb committed rFBHGX92460c4c947d: basestore: remove unnecessary files with suffix '_old' during clean up.
basestore: remove unnecessary files with suffix '_old' during clean up
Dec 4 2017, 9:00 PM
singhsrb committed rFBHGX3ba13425adb0: basestore: use tryunlink from mercurial.util to try and remove files.
basestore: use tryunlink from mercurial.util to try and remove files
Dec 4 2017, 9:00 PM
singhsrb created D1588: incremental-repack: do not repack files over a configuration based size.
Dec 4 2017, 8:11 PM
singhsrb updated the diff for D1586: incremental-repack: adding documentation for some repacking options.
Dec 4 2017, 8:11 PM
singhsrb updated the diff for D1587: incremental-repack: refactor to facilitate easier options manipulation.
Dec 4 2017, 8:11 PM
singhsrb updated the diff for D1587: incremental-repack: refactor to facilitate easier options manipulation.
Dec 4 2017, 6:34 PM
singhsrb created D1586: incremental-repack: adding documentation for some repacking options.
Dec 4 2017, 4:35 PM
singhsrb created D1587: incremental-repack: refactor to facilitate easier options manipulation.
Dec 4 2017, 4:35 PM

Dec 1 2017

singhsrb abandoned D1484: incremental-repack: prefer small packs across generations over largest ones.

Had an offline discussion with @durham regarding this. He pointed out enough cases where this might not be a better idea over the current algorithm. Therefore, abandoning it to clear the queue. There will be some follow up commits which incorporate the good ideas from this commit.

Dec 1 2017, 9:05 PM
singhsrb added inline comments to D1484: incremental-repack: prefer small packs across generations over largest ones.
Dec 1 2017, 7:56 PM
singhsrb updated the diff for D1484: incremental-repack: prefer small packs across generations over largest ones.
Dec 1 2017, 7:49 PM
singhsrb accepted D1572: infinitepush: be compatible with upstream change.
Dec 1 2017, 4:28 PM
singhsrb updated the diff for D1569: basestore: remove unnecessary files with suffix '_old' during clean up.
Dec 1 2017, 2:37 PM
singhsrb created D1571: basestore: use tryunlink from mercurial.util to try and remove files.
Dec 1 2017, 2:34 PM
singhsrb added inline comments to D1569: basestore: remove unnecessary files with suffix '_old' during clean up.
Dec 1 2017, 2:19 PM
singhsrb accepted D1570: repack: support --packsonly --background.
Dec 1 2017, 2:12 PM
singhsrb created D1569: basestore: remove unnecessary files with suffix '_old' during clean up.
Dec 1 2017, 1:28 PM

Nov 30 2017

singhsrb closed D1561: basestore: remove empty directories recursively during clean up.
Nov 30 2017, 8:23 PM
singhsrb committed rFBHGX545ee92bfc45: basestore: remove empty directories recursively during clean up.
basestore: remove empty directories recursively during clean up
Nov 30 2017, 8:22 PM
singhsrb added inline comments to D1484: incremental-repack: prefer small packs across generations over largest ones.
Nov 30 2017, 4:37 PM
singhsrb updated the diff for D1484: incremental-repack: prefer small packs across generations over largest ones.
Nov 30 2017, 4:23 PM
singhsrb added a comment to D1561: basestore: remove empty directories recursively during clean up.

@quark I think I have addressed most of your concerns. Also, the earlier code was potentially trying to remove empty directories from the whole cache. So, I have changed it a bit to only care for the repository in consideration.

Nov 30 2017, 2:00 PM
singhsrb updated the diff for D1561: basestore: remove empty directories recursively during clean up.
Nov 30 2017, 1:57 PM

Nov 29 2017

singhsrb added inline comments to D1561: basestore: remove empty directories recursively during clean up.
Nov 29 2017, 8:54 PM
singhsrb updated the diff for D1561: basestore: remove empty directories recursively during clean up.
Nov 29 2017, 8:54 PM
singhsrb updated the test plan for D1561: basestore: remove empty directories recursively during clean up.
Nov 29 2017, 7:28 PM
singhsrb updated the test plan for D1561: basestore: remove empty directories recursively during clean up.
Nov 29 2017, 7:25 PM
singhsrb updated the summary of D1561: basestore: remove empty directories recursively during clean up.
Nov 29 2017, 7:15 PM
singhsrb created D1561: basestore: remove empty directories recursively during clean up.
Nov 29 2017, 7:11 PM
singhsrb added inline comments to D1523: repack: delete excessively big packfiles before incremental repack.
Nov 29 2017, 4:36 PM
singhsrb accepted D1541: treemanifest: fail gracefully in prefetch if no remote is configured.
Nov 29 2017, 4:11 PM

Nov 22 2017

singhsrb added a comment to D1484: incremental-repack: prefer small packs across generations over largest ones.

I think the biggest diversion with this commit from the current behavior is that the repack incremental can leave 2 huge pack files even after running infinite times with the current repacksizelimit (which is 100MB) and generations (which are 1GB, 100MB, 1MB) for the datapacks. For example, say you already had two pack files, A with size 45 GB and B with size 2 GB. These files would never be packed together until there is another pack file C which has size > 1GB. When C is eventually generated, it will repack with the B (because its the smaller between A and B) to result in D and leave A untouched. Now, A and D will never be repacked till there is another pack E which has size > 1GB and so on.

Nov 22 2017, 2:15 PM

Nov 21 2017

singhsrb closed D1489: lint-checks: remove unused imports and unnecessary space.
Nov 21 2017, 10:20 PM
singhsrb committed rFBHGXe885e07707b7: lint-checks: remove unused imports and unnecessary space.
lint-checks: remove unused imports and unnecessary space
Nov 21 2017, 10:20 PM
singhsrb created D1489: lint-checks: remove unused imports and unnecessary space.
Nov 21 2017, 9:24 PM
singhsrb updated the diff for D1484: incremental-repack: prefer small packs across generations over largest ones.
Nov 21 2017, 9:10 PM
singhsrb updated the diff for D1484: incremental-repack: prefer small packs across generations over largest ones.
Nov 21 2017, 4:18 PM
singhsrb planned changes to D1484: incremental-repack: prefer small packs across generations over largest ones.

Actually, its best to avoid the highest generation if we are going across generations. So, this needs a bit of change.

Nov 21 2017, 3:07 PM
singhsrb created D1484: incremental-repack: prefer small packs across generations over largest ones.
Nov 21 2017, 2:59 PM
singhsrb closed D1461: prefetch: do not attempt to prefetch trees for non-public commits.
Nov 21 2017, 1:55 AM
singhsrb committed rFBHGX64bdb2fbbd61: prefetch: do not attempt to prefetch trees for non-public commits.
prefetch: do not attempt to prefetch trees for non-public commits
Nov 21 2017, 1:55 AM

Nov 20 2017

singhsrb accepted D1464: sparse: try reading from working copy first.

Cool!

Nov 20 2017, 4:21 PM
singhsrb created D1461: prefetch: do not attempt to prefetch trees for non-public commits.
Nov 20 2017, 1:44 PM