Conversation
… new parser class, tests rework WIP
- Integrated `prepare_text_for_dbml` into `default_to_str` for consistent handling of special characters in DBML column strings. - Updated the test suite to include a new test case for handling binary string input (`b'0'`).
…default-escapse Fix bug in DBML column default rendering with single quotes
…brackets fix: line breaks in column and index options are allowed (Vanderhoof#48)
…tion fix: table elements may go in any order (Vanderhoof#49)
feat: TableGroup notes support
Feat/table group color
fix: quote tablegroup and project name on dbml render Vanderhoof#46
Develop into master
…_other_index_types feat(indexes): support all Postgresql index types
|
Actually imports organization should not affect the performance. Separate-line imports was a design choice when I was beginning the project, not so sure about it right now, because IDEs like PyCharm manage imports pretty good for you and separate-line imports do not really make much sense at this point. But again, this is just code organization and should not be related to performance issues |
|
Initially I also had similar thoughts. But actually I am using PyCharm, and indeed reorganization of imports did trick - code is 2-3 times faster. Maybe this is because importing not just functions, but also variable declarations multiple times that calls pp many times. |
|
That's really strange! I just tried parsing the same file with and without collapsed imports, and saw no difference (86ms with one-line imports and 87ms with collapsed) |
|
This time improvement was prior disabling unicode. Yes, it has
practically no impact after disabling unicode.
|
10 participants
Moving multiple imports from same sources into single call.
This should reduce DBML parse time more than twice.
See #59