components/lwip/lwip issue on Toitware esp-idf fork #45
Labels
bug
Something isn't working
Comments
|
This works for me. I ran the same commands as your dump and it worked. You have a strange line: In the middle of the text "50148/50148" there is an inserted text "Cloning into 'components/lwip/lwip'..." It's as if there are two clones going on at once, or you have a background program and a foreground program trying to do the same thing twice? Does it work to delete everything and start over? |
|
You're right, that looks really strange. I did it a couple fo times, but have nothing of any sorts, that could trigger anything running in parallel. Although I pasted the 5 commands into the shell in one go, but that usually works fine. This is on a spare machine running Ubuntu 16.04 (EOL'ed) with git version 2.7.4. On Ubuntu 20.04 it also works for me, but when I inspect the repo I can't locate the 0263 commit that the containing repo wants to checkout, consistent with the error message. I just redid it from scratch, manually pasting the lines one-by-one and I get the same result., don't see any sign of it being accessed multiple time. My best guess ATM is that this is an "old git" artefact; |
|
Should we close this is as 'working as intended' for now, @mikini? |
|
I've been looking a bit into this as I think I see an inconsistency on the Toitware fork, that is somehow mitigated by more recent git clients. The fork is actually referencing a commit for the submodule which is part of another repository than the repository indicated by .gitmodules (Toitware vs Espressif). Whether this can lead to other problems than the one I experienced using an old git client is hard for me to predict, but to me a non-ambiguous description of the repositories' relationship is preferable. I'll try to substantiate this with some facts, everything done on command line is on Ubuntu 20.04/git 2.25.1. components/lwip/lwip submodule is cloned from espressif: However, the submodule reference points to commit 02638065e72ea689eb0d3509b1960b31c515616e which is not included on any branches of the espressif repository, but which has somehow been checked out as a detached HEAD by new git clients anyhow: Trying to reference this commit on a plain clone of esp-lwip will fail, but on the web interface github will show the commit but warn that it is contained on a fork: It seems that somehow the checkout of an unknown commit reference from a submodule reference will/can result in a search or traversal of forks and fetching of objects from that fork. Maybe that is why duplicated "Cloning into 'components/lwip/lwip'..." is seen when done on older git (but somehow still doesn't fetch the correct objects)? Looking at what happens during recursive submodule update and init on git 2.25.1 actually shows a subtle extra fetch being done during checkout on components/lwip/lwip (for reference complete dump below): Complete esp-idf checkout |
floitsch
added a commit
to toitware/esp-idf
that referenced
this issue
Jul 8, 2022
Newer gits seem to be clever enough to find it even if the path isn't set correctly. Fixes toitlang/toit#45
floitsch
added a commit
to toitware/esp-idf
that referenced
this issue
Jul 8, 2022
… set correctly. Fixes toitlang/toit#45
floitsch
added a commit
to toitware/esp-idf
that referenced
this issue
Jul 8, 2022
Newer gits seem to be clever enough to find it even if the path isn't set correctly. Fixes toitlang/toit#45
This was referenced Jul 8, 2022
|
We believe this has been fixed as part of upgrading to ESP-IDF v4.4. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
There seems to be a problem with the registered hash of the components/lwip/lwip submodule in Toitware esp-idf fork.
Relevant excerpt when following README.md procedure:
Complete dump:
The text was updated successfully, but these errors were encountered: