search

iterative / katacoda-scenarios

Interactive Katacoda Scenarios
https://www.katacoda.com/dvc
2 stars 11 forks source link

Update Scenarios for DVC 2.0 #20

Closed iesahin closed 2 years ago

iesahin commented 3 years ago

Currently there are a few issues in documentation repo for katacoda scenarios. Basically these are not running because of a required system upgrade to Python 3.8.

Update 2021-03-08

iesahin commented 3 years ago

Could you arrange the permissions/membership so that I can take over updating katacoda? @shcheklein

Thank you.

shcheklein commented 3 years ago

@iesahin I've added you to the repo. You can now do PRs and they are merged it will deployed I think. Feel free to merge them as you go, unless it's some substantial change to the content and you feel that review is needed

iesahin commented 3 years ago

Thank you.

iesahin commented 3 years ago

I have updated the versioning part.

iesahin commented 3 years ago

Changed the name because most of the parts require an update for 1.0 as well. After the release we can update the commands and breaking changes for 2.0

jorgeorpinel commented 3 years ago

Hi @iesahin 👋. I noticed you started committing directly to master recently (80 or so commits: https://github.com/iterative/katacoda-scenarios/compare/0f302dd...2993d87 up til now). It would be a better practice (and more readable repo history) to use PRs (from branches directly on this repo, no fork needed) even if you have master write access. Thanks!

image

See https://github.com/iterative/katacoda-scenarios/network

jorgeorpinel commented 3 years ago

A subtask in DVC 2.0 doc update: iterative/dvc.org#2026

Seems like that should be a separate issue in this repo? I'd also add checkboxes to the description above for each Get Started chapter:

jorgeorpinel commented 3 years ago

Question: what's the purpose of updating scenarios but keep using 1.0? Should we focus on updating them with 2.0 along the way? Per https://github.com/iterative/dvc.org/issues/2026

iesahin commented 3 years ago

Hi @iesahin 👋. I noticed you started committing directly to master recently (80 or so commits: 0f302dd...2993d87 up til now). It would be a better practice (and more readable repo history) to use PRs (from branches directly on this repo, no fork needed) even if you have master write access. Thanks!

@jorgeorpinel Hi. 👋🏼 I began writing with PRs but the iteration takes 2x time that way. You need to merge to master before seeing the effects in katacoda. There are no prior tests or whatnot. 80 commits mean ~80 PRs. I agree that it would be better documentation-wise but the workflow needs updates to the master for each commit, either merging from PR or by direct commits.

I thought establishing a basic structure first and iterating with PRs is more reasonable.

Question: what's the purpose of updating scenarios but keep using 1.0? Should we focus on updating them with 2.0 along the way? Per iterative/dvc.org#2026

The initial title for this issue had DVC 2.0. Then, as a stable release was needed to install with snap orapt, I decided to stick to 1.x and changed the title. Then we talked with @shcheklein and he said it's better to just move ahead to 2.0. I updated the installation scripts to use pip --pre instead of snap and wrote stages, pipelines, params etc. for 2.0 but forgot to update this issue's title. Currently, the scenarios don't (or shouldn't) contain any reference to dvc run and we'll have an Experiments scenario this week. I will go back to earlier scenarios and change the commands to 2.0 versions.

Almost all the content and scripts in scenarios required update. I didn't stick to the issues. (e.g. the scenarios weren't able to run because the containers were not able to install DVC in the first place.) I'll just review and check if the issues are done, move them here and see whether we have some missing items.

Thank you for reviewing all. 🙏🏼 Muchas gracias. I'll update the scenarios in PRs.

jorgeorpinel commented 3 years ago

I began writing with PRs but the iteration takes 2x time that way.

Shouldn't take that much longer, maybe 10-20% extra time? But it's worth it. Without branching or PRs it's pretty hard to go back and find where/why things were updated. For example your recent PRs are very clear e.g. #31 fixes up the language in the GS/Init scenario. We can go there and comment on any further/pending issues related to that, etc. (divide and conquer). BTW we can make reviews before you merge PRs if you want, you just need to request a review before merging.

Look how much better organized the Git history is now, with branches (you can tell where they start from) that have all the commits, while master only has merges:

image

Finally, this kind of an industry standard for collaborative/public repos. Further reading if you're interested: https://www.gitkraken.com/learn/git/best-practices/git-branch-strategy.

jorgeorpinel commented 3 years ago

we talked with @shcheklein and he said it's better to just move ahead to 2.0... I will go back to earlier scenarios and change the commands to 2.0 versions.

Sounds good!

Almost all the content and scripts in scenarios required update.

Yes they were probably made for dvc 0.x still! At least now they run and have decent, updated explanations, which is the most important step 👍

Looking forward to the 2.0 updates (again, feel free to request in-PR reviews before merging) @iesahin. Thanks

jorgeorpinel commented 3 years ago

See also https://github.com/iterative/dvc.org/issues?q=is%3Aissue+is%3Aopen+katacoda

iesahin commented 2 years ago

I think the portions related to katacoda in iterative/dvc.org#1943 are resolved. We can close this.