-
Notifications
You must be signed in to change notification settings - Fork 82
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Silabs #274
Silabs #274
Conversation
tecimovic
commented
Oct 27, 2021
•
edited
Loading
edited
- Performance enhancements: this shaves of about 25% of generation time for the ZigbeePro stack.
- Add cascading deletes to many tables that didn't have this set correctly
.dbGet(db, `${ATOMIC_QUERY} WHERE PACKAGE_REF = ? AND UPPER(NAME) = ?`, [ | ||
packageId, | ||
name == null ? name : name.toUpperCase(), |
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.
So in selectAtomicSizeFromType
we lowercase the incoming string, and rely on the stored value being lowercase. But here we are uppercasing both. Why the inconsistency?
I guess this is pre-existing, but could we then at least not have the new queries we are adding here do the uppercasing thing?
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.
It's a valid point. I simply preserved the previous muddle and added the cache. These queries are not new, they were simply isolated out of query-zcl.js.
Let me think about this a bit... I don't like any of this case-insensitivity.... Pretty much any language environment zap is targeting has case-sensitive type namespaces, and if someone wants to generate COBOL stuff from zap, I guess I'll just open a bottle of beer and cheer for them!
So let me spend a bit of thinking time around how to clean this up for good.
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.
So to be clear, I am OK with keeping the existing queries; the question is what to do for the caching bits.
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.
Caching bits are initially a bit more promiscuous because we had separate queries that in one case were, and in other case were not case sensitive. What I want to do:
- clean this whole mess up, and make it strict in ALL cases, after I figure out what the root problem is and probably end up cleaning up the XML files.
- meanwhile, use these in a maximally promiscuous mode, matching non-case sensitive, since any intermediate solution feels just more messy. Hence the cache being even more promiscuous.
…k out the source for executable testing.
JIRA: EMZIGBEE-8369
JIRA: EMZIGBEE-8369
Rebased against master before merge. |
project-chip/zap#274 should have fixed the problem that caused them to be commented out. Fixes project-chip#11056 Fixes project-chip#11057
project-chip/zap#274 should have fixed the problem that caused them to be commented out. Fixes #11056 Fixes #11057
project-chip/zap#274 should have fixed the problem that caused them to be commented out. Fixes project-chip#11056 Fixes project-chip#11057