Plots2: Wiki editing seems to be broken

[jywarren]

Hi all -- we seem to have a new bug with the wiki editing system; perhaps related to some recent changes to https://github.com/publiclab/plots2/blob/master/app/controllers/wiki_controller.rb ?

[siaw23]

Can you please get this for us? https://github.com/publiclab/plots2/issues/4903 ? That way we don't have to wait for someone to report, it show up automatically what exactly is happening and how it happened then we can all look together and fix it. It's helpful I think.

[grvsachdeva]

Hey @jywarren, raising PR for this one. Thanks!

[siaw23]

Reason for my post is I can't reproduce locally. Editing works fine locally.

[grvsachdeva]

I tried editing the wiki locally and got error.

[siaw23]

As which user? normal or admin? @gauravano

[grvsachdeva]

As moderator. Actually, here- https://github.com/publiclab/plots2/commit/40b6624af14b10027196b049e3a1ed8e11dd287e#diff-528416b5954756daa0b141666006dbcb during clean up there was a slight change in order of statements, due to which node.vid was getting null but as per schema it can't be null so 500 was thrown

[siaw23]

Yeah I missed that. Thanks for catching it. My main concern though is why didn't the test catch it though.

[grvsachdeva]

Yes, it's really shocking that Travis wasn't able to catch this

On Fri 8 Mar, 2019, 2:39 AM Emmanuel Hayford, [e-mail]
wrote:

Yeah I missed that. Thanks for catching it. My main concern though is why
didn't the test catch it though.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/publiclab/plots2/issues/4941#issuecomment-470695301,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AT6S9hX_jJHRXZYip1t_UrCeO951mihYks5vUYAigaJpZM4bj-Ql
.

[jywarren]

@siaw23 re: #4903 i think it's a great idea. I am scrambling today to keep up so let me do my best to look at it. Thanks!

[jywarren]

@gauravano would you mind looking through the logs to see how this somehow mightn't have been caught? Are we really catching the success in the test? Are we doing it to an existing wiki page, anything maybe about the context or parameters or db state that could conceivably be different? Could we run the update twice in the test?

[grvsachdeva]

@jywarren will go through the logs and try to find the reasons.

[grvsachdeva]

Build logs were not showing enough info so I restarted the build - https://travis-ci.org/publiclab/plots2/builds/501925116?utm_source=github_status&utm_medium=notification

[grvsachdeva]

It failed as the branch - https://github.com/publiclab/plots2/pull/4902 is deleted

[grvsachdeva]

@jywarren I can't see any way we can see detailed build logs for https://github.com/publiclab/plots2/pull/4902 via Travis. Any idea?

[jywarren]

All i can think of is going through the builds on the master branch from more than 12 days ago.... https://travis-ci.org/publiclab/plots2/builds

[grvsachdeva]

I went through each commit of #4902 and found that the travis build didn't run when the bug introducing commit is made mistakenly :point_down:

[grvsachdeva]

As seen in the above Gif the small refactor commit doesn't show any Travis build. But, I think subsequent builds should've found the bug but don't know why no test was able to find it.