Decommisson Addons

[maxime-rainville]

We don't currently have the time to properly support Addons. The site currently has several broken features, report incorrect information for many module and is running outdated dependencies. The queued jobs running the site appear to be mostly broken.

We are in the process of a DotOrg rebuild, which may include equivalent functionality. In the mean time, the current addon website is actively armful.

Alternative

VETOED
Reduce complexity and storage by removing unnecessary stuff and only storing modules with versions compatible with supported versions of Silverstripe CMS.

• Back to fundamentals: Remove anything that isn't essential addons.silverstripe.org#288

Note

We should probably draft and approve the copy for the page we'll be redirecting to before doing any of the decommission steps.

Acceptance criteria

• addons.silverstripe.org redirect to a page on DotOrg.
• The page explains why the site has been decommissioned and provide alternative means of searching for Silverstripe CMS module.
• Addon site is decommissioned.
• Addon stack is decommission.
• This repo is archived with a notice on the readme.
• Addons is removed from the Silverstripe asset register (Max will do)
• Removed from zenhub
• Removed from elvis (was never in it)
• Removed from rhino (was never in it)

PRs

• DOC Archive repo addons.silverstripe.org#289
• https://github.com/silverstripeltd/silverstripe.org/pull/185
• DOC Remove references to old Addons website developer-docs#389

Draft page on silverstripe.org (for peer review)

• Go to the admin section of silverstripe.org and search for "Composer installed modules"

[GuySartorelli]

provide alternative means of accessing this content.

Note that as discussed in https://github.com/silverstripeltd/product-issues/issues/818:

addons provides the ability to search for modules which are compatible which specific versions of Silverstripe CMS. This isn't possible via packagist website.

We can't provide alternative means of accessing that content without implementing it somewhere. A solution was suggested in that issue:

it would probably be easier and more beneficial to contribute that functionality to packagist than it would to keep or rebuild the addons site.

[GuySartorelli]

I've updated my thinking on this:

We should check the analytics to see how used this is. I can't see an easy way to add the "compatible with CMS X.Y" functionality which addons is really useful for in packagist directly, so it may be better to instead pare this back to essentially just that most basic functionality - upgrade to framework 4.13+, remove all the jank, remove the rating system, remove a lot of the filters and categories - and just have a simple "search for modules/themes/recipes with optional compatibility filters"

[emteknetnz]

I've created a draft page on silverstripe.org that has some minimal text about addons being retired and search on packagist instead. I really don't think we need to go in to any great explanation here, I've provide a link to search for "silverstripe" on packagist. The silverstripe.org rebuild should provide a more better way to find relevant silverstripe modules.

If you feel differently then feel free to provide new copy.

To find the page go to the admin section of silverstripe.org and search for "Composer installed modules". Once this has bee published we can move forward with getting the stack removed and the domain redirected to the published page.

[GuySartorelli]

I think instead of adding a whole new page, we should just replace the search box and browse button on https://www.silverstripe.org/software/addons with the new copy.

I also recommend this change to the copy:

- If you are looking for composer installed modules for Silverstripe CMS then you can search for them directly on Packagist using [this link](https://packagist.org/packages/silverstripe/admin?query=silverstripe).
+ If you are looking for composer installed modules for Silverstripe CMS then you can search for them [directly on Packagist](https://packagist.org/packages/silverstripe/admin?query=silverstripe).

Plus, of course, make it an actual link since markdown doesn't get evaluated.

[michalkleiner]

Is it actively harmful? Do people report that?
Can we just wait until the DotOrg rebuild is done and make a call after that?

[maxime-rainville]

For full context, we're in the process of rebuilding the DotOrg website. This feature will be reimplemented on the new website in some shape or form.

Analytics on the site is broken so I don't know how many people are actually using it. A lot of the features either don't work or give you misleading information e.g.: We don't use Travis for CI any more, but the Addons site is still looking for it as a marker of quality.

We got a warning that the broken queued jobs were consuming too much resources and would increased our hosting bill. We concluded fixing the broken jobs could be achieved by pulling back on the Addons feature set ... but that it would take at least a few days of dev time.

Given that the feature is meant to be reimplemented in a new site in a few month (:crossed_fingers:) and I didn't have clear indications that the site was actively useful to many people, I concluded it wasn't worth spending a few days of dev time fixing this.

[GuySartorelli]

(a few months is probably overly optimistic fyi)

[michalkleiner]

How much effort do you think it would be to throw a temporary CSS over things to hide what's misleading (or remove it from the template/components) and disable the queued jobs (what are they doing, do we know?) and trigger whatever needs triggered e.g. once a day or once a week from e.g. pingdom or such?

Just trying to see if there's a simple middle ground way where we won't shut this down for a a few months to a year+ without a replacement.

I'd be willing to spend a bit of time to look into this, I find the search there quite useful even without all the metrics and rating based on the obsolete module quality criteria.

I'm not saying we should keep this around forever if we think it's past its death, but it feels like a one of the "we don't know, we don't like it, we don't have data, so let's close it" decisions.

Obviously who funds it has the last say here, but my gut feel is it might be jumping the gun too quickly. For sure I don't have the visibility on what other priorities is this competing with etc.

Note - this is nothing against the squad obviously, I might be missing internal info/context or missed discussions on this etc., just trying to understand it more and not throw the baby out that easily.