You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When overwriting a file on prefix updates it might try to overwrite a file that is still in use. The workaround is to stop all usages of the prefix before an update.
Where is this important
A common use-case is in a pixi project that is activated/used in an IDE. These IDE's might be using the Python Interpreter for the language server.
The error you'll get on a package update will then look like this:
Which is hard to read and UX unfriendly. A way to avoid this error entirely would be an amazing improvement for the UX.
Solution
This has been done before in mamba, by renaming it to a xxx.conda_trash. Then another run could look for all the trash files and actually remove them.
Initial issue: Sometimes, removing a file that was (just) in use fails on Windows. I think it's good practice to add a retry after a few ms or more (we also hit something like that in pixi and fixed it with a sleep: prefix-dev/pixi@0ee497f
The text was updated successfully, but these errors were encountered:
If I were to take a stab at this, would you prefer to have a design discussion here first or that I just make an attempt at it and then we can have a discussion on the preferred solution in the pull request?
When overwriting a file on prefix updates it might try to overwrite a file that is still in use. The workaround is to stop all usages of the prefix before an update.
Where is this important
A common use-case is in a pixi project that is activated/used in an IDE. These IDE's might be using the Python Interpreter for the language server.
The error you'll get on a package update will then look like this:
Which is hard to read and UX unfriendly. A way to avoid this error entirely would be an amazing improvement for the UX.
Solution
This has been done before in
mamba
, by renaming it to axxx.conda_trash
. Then another run could look for all the trash files and actually remove them.Initial issue:
Sometimes, removing a file that was (just) in use fails on Windows. I think it's good practice to add a retry after a few ms or more (we also hit something like that in pixi and fixed it with a sleep: prefix-dev/pixi@0ee497f
The text was updated successfully, but these errors were encountered: