Recommend >= 0.2.0 of torch-tb-profiler
#16440
Conversation
|
What's the difference between this and #16361? |
|
@brettcannon #16361 will be updated to simply port this change from release into main (I intentionally left it as draft since it shouldn't be reviewed or merged yet). Apologies for the confusion. |
| @@ -194,14 +197,14 @@ export class TensorBoardSession { | |||
| // First see what dependencies we're missing | |||
| let [tensorboardInstallStatus, profilerPluginInstallStatus] = await Promise.all([ | |||
| this.installer.isProductVersionCompatible(Product.tensorboard, '>= 2.4.1', interpreter), | |||
| this.installer.isInstalled(Product.torchProfilerImportName, interpreter), | |||
| this.installer.isProductVersionCompatible(Product.torchProfilerImportName, '>= 0.2.0', interpreter), | |||
There was a problem hiding this comment.
This is minor, but these two '>= 0.2.0' and '>= 2.4.1' strings look like they would be better as a constant given that you use them in two places already. Feels a bit too easy to change one location and forget the other.
There was a problem hiding this comment.
@joyceerhl Ahh, I see this is direct to release. I don't want you to have to retest for this. So maybe we could clean this up in a later PR to main? Changing to approve, nothing wrong functionally as is.
|
@joyceerhl It seems smoke tests are failing after merging this https://github.com/microsoft/vscode-python/runs/2826444216?check_suite_focus=true. Can you look into it? |
|
@karrtikr Those failures are due to the smoke tests in the Python extension requiring an update--the Jupyter extension recently deprecated a setting, and the Python extension tests still reference that setting. The failure is unrelated to this change. Do you want me to submit a PR to release to update the smoke tests? |
|
I see, I assume they're still specific to smoke tests related to data science. If that's the case please submit a PR. |
* Recommend >= 0.2.0 of `torch-tb-profiler` (#16440) * Recommend >= 0.2.0 of torch-tb-profiler * Update CHANGELOG.md and delete news * Extract semver requirements into constants * Update CHANGELOG.md Co-authored-by: Kim-Adeline Miguel <[email protected]> Co-authored-by: Kim-Adeline Miguel <[email protected]>
* Recommend >= 0.2.0 of `torch-tb-profiler` (#16440) * Recommend >= 0.2.0 of torch-tb-profiler * Update CHANGELOG.md and delete news * Prep for trusted workspaces work (#16485) * Implement python.defaultInterpreterPath in workspace scope * Move DeprecatePythonPath experiment to new framework and retire old framework * Remove execution isolation script * Do not delete old user setting while migrating interpreter settings * Modify the python path deprecation prompt * Add deprecate pythonPath doc in package.json * Add default Interpreter path entry at the bottom of the interpreter list * Update VSCode engine to 1.57 * Restore Deprecate PythonPath experiment for compatibility * Remove environments security related to discovery (#17) * Remove environments security related to autoselection (#18) * Fix ESLint errors related to security undo work * Add experiment to package.json * Edit changelog to include enhancement details * Delete news entry * Update release candidate for release (#16490) * Update change log * Update version * Update TPN * Remove deprecated setting usage in DS smoke test (#16491) Co-authored-by: Joyce Er <[email protected]> Co-authored-by: Kartik Raj <[email protected]>
For microsoft/vscode-jupyter#14878
This is going directly into the release branch for now, because
torch-tb-profileris slated to release 0.2.0 (which contains jump to source support) June 15. We can't merge into main before that as it'll cause users in Insiders to get prompted to upgrade theirtorch-tb-profilermodule install when the upgrade is not yet available on PyPI.I tested this PR, and jump to source support, against their release candidate (
pip install torch-tb-profiler==0.2.0rc3).I'll port this change back to main when 0.2.0 is released.