Squash deployed gh-pages, relying on this repo for history #658
Conversation
|
|
|
💩 |
|
@jni looks like this obliterated everything but |
|
Looks like |
|
Sorry that i missed this PR btw. My notifications have been overwhelmed with spam from my own docs fork. |
|
um. Shit. |
|
ok, the minimal damage control rn is to redirect napari.org to dev. Let me add that. |
|
ok I have history locally up to 0.5.4. |
|
My gif was so appropriate |
|
Ooof, can't even revert it because history is gone. 😬 |
Using the latest commit is what we want if the latest commit contains all the folders, which I thought it did. Unfortunately this clobbers the whole history so there is no going back. However we may only lose 0.5.5 and 0.5.6 in the worst case. I feel very dumb for not doing git clone before this. 🤦 |
|
ok let me first revert this so we can modify things going forward without them getting clobbered. |
…apari#658)" This reverts commit 5fc3cd7. See discussion starting at napari#658 (comment)
|
See #663. Could I have a quick merge please? 🙏 |
|
@jni approved, but i'm about to hit the hay -- sorry! 😞 |
|
haha better than last time when I clobbered the history and then myself hit the hay 🤣 Have a good night! This may save us, will explore now: |
|
sleep tight! 😂 |
|
This seems useful -- or is this that backup you mentioned @jni? It has 0.5.6 though https://github.com/napari/napari.github.io/tree/gh-pages-backup-2025-04-08 |
|
Oh sweet bejeebus. https://github.com/napari/napari.github.io/tree/gh-pages-backup-2025-04-08 😂 for the record, I followed the instructions from the medium post, but the curl POST command wouldn't work (probably because of 2FA on my account), but I followed the instructions to the docs and used the GH CLI instructions: https://docs.github.com/en/rest/git/refs?apiVersion=2022-11-28#create-a-reference |
|
I just made that right now @TimMonko thanks to the above links 😅 😅 😅 |
)" (#663) This reverts commit 5fc3cd7. See discussion starting at #658 (comment)
|
Oh wonderful. I am currently cloning it too, just in case. I won't know what to do with it should things go awry but I'm here for you (while fixing my 6th bike flat of the spring gahhh) |
|
actually, somewhat interestingly, I just saw that the gh-pages deployment action gives you an artifact: So grabbing from that would have been another option. 😅 Good to know for the future. https://github.com/napari/napari.github.io/actions/runs/14371590037 Anyway, we are back in business!!!! 😅 Everyone can go back to their normal levels of anxiety now. 😂 |
|
I went a bit too far back on the restore (didn't want to waste time restoring the same commit 😅) so we've lost Tim in the about page. Small price to pay. 😜 No, in seriousness, now that #663 is merged, the very next deploy from here should fix that. 😅 |
|
Lol I'm such a dingbat I was already using the docs for the past few minutes without realizing it ever went down 🙃 |
|
The lesson here is that sometimes you have to be the villain in order to become the hero |
|
Oh cheese I'm so sorry I missed all the fun here 😆 Just trying to catch up on notifications... |
In #621 and elsewhere, we are discussing the issue of autogenerating
screenshots in the docs. @Czaki brought up the issue that git is very
bad at handling binary blobs, so autogenerating images each time we
build the docs would rapidly blow up the repo clone size.
I think it's clear that we do want to autogenerate images, so the
cleanest solution in my opinion is to always squash when pushing to
our gh-pages site (napari.github.io). That history is not particularly
valuable, and indeed we have squashed it once before when it was getting
far too big to work with. Doing it automatically is actually an
improvement.
This PR achieves that by using the
force_orphanoption in thepeaceiris/actions-gh-pages action to always squash the edit history
before pushing to the gh-pages branch.
References
Zulip discussion:
https://napari.zulipchat.com/#narrow/channel/298358-working-group-documentation/topic/Meeting.202025-03-13.2F14/near/505543160
force_orphandocumentation:https://github.com/peaceiris/actions-gh-pages?tab=readme-ov-file#%EF%B8%8F-force-orphan-force_orphan