Make GTO faster #163
Make GTO faster #163
Conversation
| } | ||
|
|
||
| def check_ref(self, ref: str): |
There was a problem hiding this comment.
@amritghimire, this is not ideal, but if you create GitRegistry, you can use this function body as an example on how to cache state. Looking for better solutions.
There was a problem hiding this comment.
This could work. A similar approach for get_stages will also be great. We do checkout to a commit for each get_stages call. But if get_stages use the head when creating GitRegistry, having ref there as well would be better.
There was a problem hiding this comment.
Why do you call get_stages each time? What stages user sees in MR should depend on config HEAD only. So it doesn't matter what is there in all other commits.
There was a problem hiding this comment.
I.e. you should only call $ gto stages when HEAD of default branch gets updated.
There was a problem hiding this comment.
We need to parse everything for each commit. Because:
- We parse incrementally and not all at once
- Some commits might be removed, in this case we don't want to reparse, we want to recalc.
In perfect world we wouldn't need to create GitRegistry and manage it. Things will be cached for us on your side, not sure how hard is this to do though.
There was a problem hiding this comment.
The easiest workaround is probably providing git_registry param, i.e. if not None then use it instead of recreating each time.
Addressing findings in #156
Checked this again, it seems and got faster by 50% only. Will see what else can be done, in the future PRs.