Strange issue with double dependency detection / writing package.json twice about jitsu

[bmeck]

Can you get a screen dump of this, from what I'm seeing it is bumping the version number if it was already deployed to an unused one and only doing it once if it has not been deployed (or if it has been destroyed).

[AvianFlu]

Reproduced.

https://gist.github.com/b13811b13a019abf3e89

[bmeck]

So the issue is that it is being written twice, but it is writing 2 different package.json files.

1. Before asking the server if this is a deployable snapshot.
2. After finding it is not and bumping the version number.

The problem with not writing twice is that if we don't write through on the first one and there is a connection error you will be prompted to save after an error (odd UX). If we don't prompt on the second we might end up w/ some UX issues from people saying it is not writing what it prompted it would. We could make version bumping not prompt the user and instead just write?

Also, adding a message about the version number being bumped a bit more explicitly.

[indexzero]

This is an order-of-operations problem in the jitsu deploy command: https://github.com/nodejitsu/jitsu/blob/master/lib/jitsu/commands/apps.js#L99-127

Ideally the workflow should be:

1. jitsu.package.read
2. If no package.json, prompt for app name and then attempt jitsu.apps.view
3. If application exists, write truncated results to disk and increment version number
4. If no application exists (i.e. on 404 error), continue with the rest of jitsu.package.get.
5. Otherwise, error out of the CLI because something bad has happened (i.e. 500)

[Marak]

@bmeck - Will you be able to resolve this issue?

[bmeck]

yes i can, but indexzero put it at a low priority so I stalled

On Mon, Jul 11, 2011 at 4:26 PM, Marak
[email protected]
wrote:

@bmeck - Will you be able to resolve this issue?

Reply to this email directly or view it on GitHub:
#55 (comment)

[AvianFlu]

It looks like I'm still seeing this issue in Jitsu v0.4.0.

https://gist.github.com/1138356

[jfhbrook]

I don't believe this is still an issue. Closing.

[indexzero]

I've seen this issue recently. I will give you a repro this weekend.

[Marak]

Am not about to reproduce / haven't seen this. Closing unless someone can post reproducible steps.