fix: use lstat bigint=true for local environments to fix #1245 #1778
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -18,5 +18,8 @@ | |
| ], | ||
| "newlines-between": "always" | ||
| }] | ||
| }, | ||
| "env": { | ||
| "es2020": true | ||
| } | ||
| } | ||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,2 @@ | ||
| /* eslint-disable no-undef */ | ||
| jest.retryTimes(3, { logErrorsBeforeRetry: true }) |
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -222,6 +222,12 @@ export class FileSystem { | |
| async lstat(filename) { | ||
| try { | ||
| const stats = await this._lstat(filename) | ||
| // For non-browser (local) scenarios regular git 'add' command might be used to write the index. Therefore we need to have the exact nanoseconds (see. normalizeStats.js SecondsNanoseconds ). | ||
|
There was a problem hiding this comment.
@dead-end thats right, and is mentioned here. if you use native git the timestamp (nanoseconds) is more accurate than the timestamp ('faked' nanoseconds by using ms * 1000000) in isomorphic-git. (this is also the reason why the git index is broken after isomorphic-git tries to rebuild it, becuase it uses the 'faked' nanoseconds) Thats why we need to differentiate between browser and local. With There was a problem hiding this comment. @jayree You have written: (this is also the reason why the git index is broken after isomorphic-git tries to rebuild it, becuase it uses the 'faked' nanoseconds) Can you explain how the faked nanos lead to the broken index file? We have a lot of issues with this and I would like to understand it. There was a problem hiding this comment. @dead-end with the old behaviour nanoseconds were calculated based on From a file with The old logic fixing the #1245 compared |
||
| if (!process.browser) { | ||
| const statsNs = await this._lstat(filename, { bigint: true }) | ||
| stats.mtimeNs = statsNs.mtimeNs | ||
| stats.ctimeNs = statsNs.ctimeNs | ||
| } | ||
| return stats | ||
| } catch (err) { | ||
| if (err.code === 'ENOENT') { | ||
|
|
||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@dead-end We can try to enhance this one but I don't know if we can hardcode timestamps to test them the way you want.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If you create a repo with git, then this should have an index file with the timestamps. Call
GitIndex.fromBufferto get the index content and then call lstat for each entry and compare the values. I think this should be sufficient.There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@dead-end tried it with this approach: e6c277a but as 'nano.txt' is checked out on the build server, the timestamp changes, so the test will always fail.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@jayree Sorry, I did not think of that. The only possibilities I can think of are calling git from node or put the nano.txt in a zipfile and unzip the file during the test run (this should preserve the timestamps). I do not know if this is possible.
@jcubic
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@dead-end I'm simply using the isomorphic-git repo now