Open silverwind opened 2 years ago
This is almost certainly due to the current architecture of using PyMeta and compiling the grammar when importing pybars._compiler
.
I had originally had interest in rewriting the project to use a different parser (https://github.com/wbond/handlebars.py), but at this point I have too many projects to work on, so I don't foresee working on this any time soon. If there was some strong interest for this, and some corporate sponsors interested in backing the work, I might reconsider.
Actually, it is possible that the compiled grammar could be serialized into a file on each release. That may give it enough of a performance boost. Perhaps that is something you could look into?
Not sure I'm motivated enough, Python is not my lingua franca. We definitely feel the pain of this import time thought, as it's used during short-lived processes, so I'm not saying I won't eventually get motivated :)
Actually, it is possible that the compiled grammar could be serialized into a file on each release.|
do you mean via the CI (Github Actions) pipeline & source code modification?
On my machine, it takes over 400ms to import
pybars
: