Optimize render_stacktrace() #1571
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
adamchainz
force-pushed
the
optimize_sql_panel_1
branch
from
4289268 to
6e4830e
Compare
This function is called by the SQL panel to render each frame of each stack trace. Previously it used a Django template, which is slow. By generating the HTML with pure Python code, we can avoid much of this overhead. I tried this change on a demo app I built that has 96 total queries, due to N+1 queries. The queries execute in 13ms due to use of SQLite, so the whole view is a negligible concern. Nearly all the view runtime is the toolbar itself. Without this change, the runtime is ~1300ms; with the change it’s ~1100ms. That's a saving of **15%**. I also checked the appearance of the generated HTML hasn’t changed.
adamchainz
force-pushed
the
optimize_sql_panel_1
branch
from
6e4830e to
3d77806
Compare
matthiask
approved these changes
Jan 12, 2022
adamchainz
commented
Jan 12, 2022
Comment on lines
-78
to
-81
| except KeyError: | ||
| # This frame doesn't have the expected format, so skip it and move | ||
| # on to the next one | ||
| continue |
There was a problem hiding this comment.
this didn't make sense - list.append cannot raise a KeyError
adamchainz
added a commit
to adamchainz/django-debug-toolbar
that referenced
this pull request
Jan 12, 2022
Add a few bits of caching: 1. Add sub-function to `parse_sql` with `lru_cache`, so that repeat calls with the same query are fast. This saves a lot of processing in N+1 situations. 2. Cache constructed filter stacks in `get_filter_stack()`. This avoids recreating all the various sqlparse objects for each query. 3. Pre-compile the simplification regex. The `re` module already uses an internal LRU cache of regexes, but this avoids recompiling if the regex ever drops out of that cache. Building on top of django-commons#1571, this takes run time for the same tested view from ~1100ms to ~950ms, another ~15% saving.
|
Thanks! |
adamchainz
added a commit
to adamchainz/django-debug-toolbar
that referenced
this pull request
Jan 12, 2022
Add a few bits of caching: 1. Add sub-function to `parse_sql` with `lru_cache`, so that repeat calls with the same query are fast. This saves a lot of processing in N+1 situations. 2. Cache constructed filter stacks in `get_filter_stack()`. This avoids recreating all the various sqlparse objects for each query. 3. Pre-compile the simplification regex. The `re` module already uses an internal LRU cache of regexes, but this avoids recompiling if the regex ever drops out of that cache. Building on top of django-commons#1571, this takes run time for the same tested view from ~1100ms to ~950ms, another ~15% saving.
matthiask
pushed a commit
that referenced
this pull request
Jan 12, 2022
Add a few bits of caching: 1. Add sub-function to `parse_sql` with `lru_cache`, so that repeat calls with the same query are fast. This saves a lot of processing in N+1 situations. 2. Cache constructed filter stacks in `get_filter_stack()`. This avoids recreating all the various sqlparse objects for each query. 3. Pre-compile the simplification regex. The `re` module already uses an internal LRU cache of regexes, but this avoids recompiling if the regex ever drops out of that cache. Building on top of #1571, this takes run time for the same tested view from ~1100ms to ~950ms, another ~15% saving.
gone
pushed a commit
to gone/django-debug-toolbar
that referenced
this pull request
Jan 16, 2022
This function is called by the SQL panel to render each frame of each stack trace. Previously it used a Django template, which is slow. By generating the HTML with pure Python code, we can avoid much of this overhead. I tried this change on a demo app I built that has 96 total queries, due to N+1 queries. The queries execute in 13ms due to use of SQLite, so the whole view is a negligible concern. Nearly all the view runtime is the toolbar itself. Without this change, the runtime is ~1300ms; with the change it’s ~1100ms. That's a saving of **15%**. I also checked the appearance of the generated HTML hasn’t changed.
gone
pushed a commit
to gone/django-debug-toolbar
that referenced
this pull request
Jan 16, 2022
Add a few bits of caching: 1. Add sub-function to `parse_sql` with `lru_cache`, so that repeat calls with the same query are fast. This saves a lot of processing in N+1 situations. 2. Cache constructed filter stacks in `get_filter_stack()`. This avoids recreating all the various sqlparse objects for each query. 3. Pre-compile the simplification regex. The `re` module already uses an internal LRU cache of regexes, but this avoids recompiling if the regex ever drops out of that cache. Building on top of django-commons#1571, this takes run time for the same tested view from ~1100ms to ~950ms, another ~15% saving.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This function is called by the SQL panel to render each frame of each stack trace.
Previously it used a Django template, which is slow. By generating the HTML with pure Python code, we can avoid much of this overhead.
I tried this change on a demo app I built that has 96 total queries, due to N+1 queries. The queries execute in 13ms due to use of SQLite, so the whole view is a negligible concern. Nearly all the view runtime is the toolbar itself. Without this change, the runtime is ~1300ms; with the change it’s ~1100ms. That's a saving of 15%.
I also checked the appearance of the generated HTML hasn’t changed.