Please, please support excluding archived repos!

[serhalp]

Hi! I'm so excited that GitHub is rebuilding and significantly improving code search! It's desperately needed.

Please, please don't rebuild a brand new code search that also does not support filtering out results from archived repos.

This missing functionality alone has easily made GitHub code search 50+% less usable than it would otherwise be for me at our org, which contains 900+ repos out of which 600+ are archived. I have to wade through so much noise while holding the context in my head about which repos are archived and therefore irrelevant. I have to let others know when they've been misled into thinking some function is used 25 times when it's actually used 3 times.

There are have been hundreds of people advocating for this for years over here: https://github.saobby.my.eu.orgmunity/t/feature-request-search-code-for-not-archived-repositories/2135/56.

If I had children, I would give up my first-born son for this. 🙏🏼

[Avivhdr]

I don't have children but I would give someone else's child for it 👍🏼

[ericLemanissier]

I have two children, but I would give 10 other children for this feature (at least)

[vronchin]

I don't have children but I would make children and give half of the total for this feature (at least)

[jbarber]

I agree this functionality would be killer. I use Sourcegraph for this exact feature.

[reevesbr]

We have thousands of repos across our organization, being able to code search search active, non-archived repos (the recently announced log4j vulnerability is the trigger for me landing here) is a must have.

[look]

Thanks for the feedback! This is on our roadmap. (The team requests please no children, though.)

[serhalp]

@look That's great news! For an added push... it was on our roadmap to jump ship to Sourcegraph for search in the coming year. With the GitHub code search rebuild announcement, we're close to abandoning that project – but without this feature (excluding archived repos from code search), we'll have to proceed anyway; with it, we'll definitely not bother with an external tool (sorry, any Sourcegraph folks reading this, you seem great).

[devinschumacher]

still not done huh?

[wagnerbsouza]

That latest log4j vulnerability just brought me here. The archived repos in the search results are "annoying".

[look]

This is now implemented!

By default, results from archived repositories are searched, but down-ranked. If a result appears only in an archived repository, you'll see it.

You can change this using is:archived and NOT is:archived. For example, if you want to find Log4J in non-archived repositories only, use:

log4j NOT is:archived

More details are available on the syntax page.

[mbargull]

@thoraage, @mikhailvartanyan, -is:archived (i.e., negate by prefixing hyphen-minus) works for me.

[mikelindsey-okta]

This is not working with -is:archived nor with NOT is:archived .. I have a search with 340 results without the exclusion. When I add the exclusion I have 356 hits

[dreusel]

Also does not work for me, and there is no mention (anymore?) on the syntax pages.

[dreusel]

Also does not work for me, and there is no mention (anymore?) on the syntax pages.

Ah, it actually does work, it just seems to take some time after archiving a repo until it actually gets excluded from search results.

[dev-willis]

I can't get is:archived or NOT is:archived to work at all but archived:false/true works.

[LivingInSyn]

I found NOT is:archived in the documentation for the current codesearch beta, but it doesn't work at all for me. Sample query I'm using is:

NOT is:archived foo

returns archived repos

[colinwm]

Did you recently archive this repo? It may take some time (~hours) for the archive status to reach the search index.

[LivingInSyn]

at least a week ago, likely much longer ago than that. This is just one example, there are many

[nkojuharov-aca]

I see the same - NOT is:archived is completely ignored - results for archived repos are shown near the top, completely obscuring results for non-archived repos.

[look]

@nkojuharov-aca sorry, we are clearly missing some updates when repos change archived state. We've prioritized a new approach to updating the index when repositories change to address the problems.

[dev-willis]

I had the same problem. I figured out that archived:false is what's needed.

[look]

@chrisrusso this is a feedback forum for Code Search and Code Navigation. Please contact customer support: https://support.github.com/contact?tags=rr-restore

[drullo]

This is only partially working for me. When searching my organization's repos, adding 'NOT is:archived' to the search does cut down the number of results, but it unfortunately still includes some archived repos.

[brett-swan-sh]

I've found in our organization that this doesn't work at all. In fact, when you codesearch for something that only exists in an archived repo, and set is: archived, the code from the archived repo doesn't show up. But when you search using NOT is:archived, the code in the archived repo does show up...

[look]

Sorry about that. This sounds like a data synchronization issue, which should be resolved the next time we rebuild the index. If you would like us to check on that, let me know the org name.

[brett-swan-sh]

Ah interesting, so there's a delay between archiving the repo and the search function reflecting archive status? Any idea what kind of delay we can expect for that?

[look]

There is a delay between a change happening and the search engine being updated, but it is also possible for a change to be processed out of order, which results in the archived status reflecting the old value. We correct this the next time a change happens, but for archived repositories, that's not usually until the next index rebuild.

[brett-swan-sh]

fair enough. Any idea what that delay looks like? 5, 15, 30 mins, 1 hour, 1 day?

[look]

The normal indexing delay is about 5 minutes. If we miss an update (like I suspect happened here), the fix won't roll out until the next rebuild. We do that about once a week or so.

[devinschumacher]

[Workaround]

Here's my workaround for all you frustrated cats (i was too till i just figured this out)

Create a new organization called yourorganization-archive

and instead of "archiving" a repo (i guess you still can do that too) you just "transfer" the repo.

boom! solved & saved :)

🍻 cheers!

now lets connect on social bc i need more developers and smart people feeds to read instead of political nonsense and camgirl spam

twitter @dvnschmhr
YT: @devinschumacher

[rzuckerm]

When is this feature going to be rolled out to GitHub Enterprise? It looks like it's been in public GitHub for well over a year. I tried this on our 3.11.2 instance, and it does not work. For example:

filename:foo

returns N matches.

filename:foo is:archived

returns the same N matches, even though I know there are archived repos.

filename:foo NOT is:archived

returns no matches.

[stormindia]

If anyone can provide an update if this has been rolled out for GitHub (or for Github enterprise in particular)? Much like the original thread creator, I would go to have a child and give it to this cause (if it serves as an inspiration 🥺 )

[LiteBrite82]

Hi @stormindia

Excellent question! Unfortunately, the answer is outside of my wheel house.

The last update from @look (July 2023):
@pascal-hofmann asked:

... I noticed that this does not work on GitHub Enterprise Server. Is there a timeline on when we can expect this feature there?

look replied:

... No timeline yet, but we are working on bringing the new code search to GitHub Enterprise Server right now.

I would, however, recommend submitting a new discussion in Code Search and Navigation addressing this as a separate topic; example: "GitHub enterprise accounts need support for excluding archived repositories in code search".

• in the body, maybe add a link to this discussion just for background information

It may seem redundant, but the new topic will help us track the community feedback more accurately and aid in collaboration.

[wyardley]

Is there a way to support -is:archived for issues / PRs, or would that have to be requested in a different place?

[LiteBrite82]

@wyardley
Issues and PRs use a different search engine than code search. That would be a great discussion post for the Projects and Issues category. https://github.com/orgs/community/discussions/categories/projects-and-issues