Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [che-dev] Stacks housekeeping

Hi Zak!

Yes, all good questions. I am sometimes puzzled when looking at a PR in che-dockerfiles repo, since merging a PR there does not mean that the image will be automatically built and pushed. Even if it is, nobody will benefit from it unless there's a stack that references this image in stacks.json in Che repository. 

Just, fyi, stacks value is almost lost with Che 7 which will allow users to pull any image as there will not be any pre-requisites for the image itself, eg JDK, dependencies for agents and language servers etc.

On Tue, Jun 5, 2018 at 4:39 PM, Zachary Sang <zacharysang@xxxxxxxxx> wrote:
Hi Eugene,

My name is Zak and I am more a user than a developer of che, although I have submitted a couple PRs to the che-dockerfiles repo which has informed the following opinion:

I think that this clean up effort could be a good opportunity to establish some sort of formal specification for what a stack's recipe should look like. Specifically, I think it would be useful to at least include a template or scaffold for recipes to be based on. I think that without this, writing Dockerfiles for recipes is more of a free-form process which results in a non-uniform set of available stacks that can be confusing to navigate (ie: developers may need to read several Stack and recipe definitions to find a stack that has what they want).

The above points fall under your mention of potential QA coverage and so I would be interested in what these future plans may entail. Let me know what you think of this!

Best,

Zak




On Tue, Jun 5, 2018, 9:17 AM Yevhen Ivantsov <yivantso@xxxxxxxxxx> wrote:
Hi Che developers!

It's been a while since what we have taken a look at https://github.com/eclipse/che/blob/master/ide/che-core-ide-stacks/src/main/resources/stacks.json. Those are Che stacks - smth that a user first sees when trying the product.

While some ready to go stacks and samples are tested every day, others seem to be abandoned. If you ask me why we have debiaa-lsp stack, I'd say - historical reasons. There are more examples. Stacks have been added but never maintained. 

As a result, we have a dozen on unnecessary stacks, or those that simply do not run on any infra.

This email is to announce an upcoming PR that:

* removes debian, debian-lsp, ubuntu and openshift default stacks. All of them were added to the product because of business or tech requirements of that time.
* moves GAE SDK to default PHP and Python images, so there's no need in additional 3-4 images
* removes Python 2.7 and PHP 5.6 images from the list, but there will be instructions in docs on how to add those back
* clears components for all stacks. It does not look like we have come to understand what components are and why we need them. Neither of the clients (UD, IDE) use them.

I am currently working on updating images, testing samples and shrinking the two jsons (samples and stacks).

Further plans may involve some QA coverage for images and sample apps we provide. The goal is to make the a perfect first time UX in Che - starting a workspace and running a sample app is typically the first thing a user does.

Let me know if you have any objections/concerns with the above plans



--

EUGENE IVANTSOV

Red Hat 

eivantsov@xxxxxxxxxx   

_______________________________________________
che-dev mailing list
che-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/che-dev

_______________________________________________
che-dev mailing list
che-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/che-dev




--

EUGENE IVANTSOV

Red Hat 

eivantsov@xxxxxxxxxx   


Back to the top