Improve RC workflow: Enhanced validation and error handling in Makefile #270
Conversation
borland667
changed the title
Pull Request Test Coverage Report for Build 9271628389Details
💛 - Coveralls |
borland667
force-pushed
the
improve-test-release
branch
from
7e2e813 to
2345fb4
Compare
borland667
requested review from
Franr,
angelofenoglio,
diego-ojeda-binbash and
exequielrafaela
borland667
changed the title
| @echo "[INFO] Fetching the latest version from PyPI." | ||
| $(eval LATEST_VERSION=$(shell curl -sL "https://pypi.org/pypi/leverage/json" | jq -r ".releases | keys[]" | $(SORT) -V | tail -n 1)) | ||
| @echo "[INFO] Latest version fetched from PyPI: $(LATEST_VERSION)" | ||
|
|
||
| # Get the latest RC version from PyPI if it exists | ||
| $(eval LATEST_RC_VERSION=$(shell curl -sL "https://pypi.org/pypi/leverage/json" | jq -r ".releases | keys[]" | grep "$(LATEST_VERSION)-rc" | $(SORT) -V | tail -n 1)) |
There was a problem hiding this comment.
At the time of writing this comment:
$ curl -sL "https://pypi.org/pypi/leverage/json" | jq -r '.releases | keys[]' | sort -V | tail -n 2
1.12.3
1.12.3rc1And so, after running these two statements
LATEST_VERSION=1.12.3rc1
LATEST_RC_VERSION=""
since rcs currently do not use the dash separator. But if we ignore this, both variables would hold the same value.
Suppose we don't have 1.12.3rc1 as a released version, we would end up with something like
LATEST_VERSION=1.12.3
LATEST_RC_VERSION=1.12.2rc3
which would create various different issues down the line like attempting to release 1.12.3rc4 without the previous 3 existing (following the logic in lines 77 to 83).
Couldn't we rely only on LATEST_VERSION?
| if [ "$(PROVIDED_RC_VERSION)" -le "$(shell echo $(LATEST_RC_VERSION) | sed 's/.*-rc//')" ]; then \ | ||
| echo "[INFO] Provided RC version is valid."; \ | ||
| else \ | ||
| echo "[ERROR] Provided RC version is invalid as it does not follow the latest RC version from PyPI. Exiting..."; \ | ||
| exit 1; \ |
There was a problem hiding this comment.
By accepting PROVIDED_RC_VERSIONs that are lesser or equal that the current latest rc version we will most likely clash with an already existing release, wouldn't we? also we wouldn't be able to progress in the rcs version if I'm not mistaken.
I think rcs should be always incrementing in version number since they are for testing purposes, what do you think about it?
|
As discussed with @angelofenoglio and @juanmatias we'll close this issue for the moment. If needed in the future we can reopen it. |
What?
bump-test-versiontarget in the Makefile to ensure proper validation of the provided release candidate (RC) versions.rc0if no RC versions exist or incrementing the latest RC version from PyPI.Why?
References
Before release
Review the checklist here