IMPORTANT NOTICE - Spring cleaning for Puppet Forg...
# forge-modules
s
IMPORTANT NOTICE - Spring cleaning for Puppet Forge: Deprecating outdated modules Now that countless Forge contributors have devoted thousands of hours to module creation, Forge has become a vast repository of reusable Puppet code. Almost every Puppet user has visited Forge at least once as part of the Puppet journey. Most users probably found the module, the usage information, or the technical inspiration they were looking for. But, with thousands of published modules, Forge has reached a point where searches are becoming more challenging, even after filters are applied, because of the large number of search results. Search results can include outdated, unmaintained modules that are typically less helpful than regularly maintained modules. To meet this challenge, the Forge team will deprecate modules that have not been updated in the last 5 years. The team will start notifying module publishers via email 6 months before the proposed deprecation date. A new badge, SCHEDULED FOR DEPRECATION, has been introduced as an informational measure. This badge is shown alongside the module in the search results and on the module page. A dynamic list of modules that are scheduled for deprecation will be maintained on this page for searches. There, you can search using the module name or the publisher's name. If a module that you maintain is on the Scheduled for deprecation list, consider releasing a new version of the module. For more information, see the FAQs. If you use a module that is scheduled for deprecation, or has been deprecated due to inactivity, you have no cause for concern. A deprecated module is not deleted. However, the module is hidden in default search results. The Puppet Forge team hopes that this effort will help maintain Forge as a reliable and updated source of software to meet community requirements. If you have questions or concerns, feel free to reach out in this Slack channel
👍 1
🎉 3
r
Would it be possible to put the 'scheduled for deprecation' in the RSS feeds for those modules
c
I think it will be important to be every bit as communicative with the community. I’m using some modules that are very old and have not been touched in years, BUT THEY WORK. I’d be careful if the modules are actually still functional, trying to deprecate them. Many of us might still be using them. Perhaps an added metric you can get is how much they’re cloned from someone’s estate, their Puppetfile, etc.
r
I think when a module gets deprecated, it still exists on the forge, it just won't show up in general searches
c
Right, you have to search manually and then click the “show all” or whatever the option is.
The main thing is that those don’t get pulled from the API, because all our collective Puppetfiles will implode if we’re sourcing them in a “forge-y” sort of way.
r
I had to fork a module last updated in 2019 and add my fork to the forge because a maintainer decided he was 'too lazy' to make any more changes, and the module wouldn't work with Puppet 8. But yeah, I get what you're saying
c
Samba, anyone?
r
lol, exactly
c
Because I did the same danged thing with Samba. Most of it around “legacy” facts.
r
windows_smb had a bunch of validate functions for the params that didn't work anymore
and I'm sure some legacy facts in there too
c
They went through a lot of trouble getting us to use
validate_*
from stdlib only to deprecate the heck out of those functions.
r
there's some modules like gcc that do nothing more than have a single package resource with the various names it presents in rpm/deb land, which might vary between the two but rarely changes, and is a dependency for other modules, and those have almost no need for updates over the years.
(it was croddy/gcc, I assume it's puppet/gcc now but haven't looked)
if modules like that got moved into the puppet namespace, probably not a big deal. but if the original owner doesn't do puppet work any more, might be difficult to get a new version out the door.
c
That’s the kind of things I’m talking about. Be careful not to deprecate modules with no activity if the reason for that inactivity is that they are done and need no changes. 😁
r
I've noticed some modules in the Puppet namespace are scheduled for deprecation as well
s
Even a fully completed module should have a metadata file that needs updated more often than every five years to show what Puppet versions and OSes are supported. It doesn't seem to be a huge lift to ask that a module get that much work done to it.
💯 1
m
i just got notified of 5 of my modules i forgot existed
😄 1
Ill probably update the metadat
s
yeah, the issue with applying this strategy is it is almost impossible to determine if a module is not updated because it does not need updates or no-one really maintains it. What we have realized through a thorough data analysis activity is that majority (and I mean more than 85%) of the modules being viewed and downloaded are fairly recently updated and hence adding a couple of steps to finding an outdated module seemed like a good tradeoff Also, we don't expect any Puppetfile implodes 😛 as we are not removing the module in any way And yes, I know there are some puppetlabs modules in there, you can call us many things, partial is not one of them 🙂
r
@spp I think it's more that a lot of the owners aren't involved in puppet anymore. And "puppet > 4" is fine for a lot of the modules!
metal 1
I didn't get any notifications, I assume mine are like 4y10m old lol
😄 1
waiting 1
r
I just raised an issue on a module just to update their metadata... then found out the guy worked at Puppet like 8 years ago... I don't think he's coming back to fix his module.
I think I'll be forking another module