User friendly search option haikudepot

That would be the plan9 way

Personally, I think adding freeform tags could be a possibility, this has been discussed bedore, I am not sure where but i think the problem was that what users might search for is not apriori knowledge, so would likely need support in the haikudepot UI to allow users to make tags there.

I disagree (that it’s more plan9 than Haiku/BeOS) - this is the way mail works currently. I think what you’re saying is that users are not technical enough these days to e.g. reconfigure Tracker windows to show different attributes etc. (as you have to with mail or contacts).

Mails have been files on unix for a long time, this is neither related nor specific to haiku. The only interesting part is that we use the attributes to add interesting info, we do not however try to expose every component as a file, in that sense we are more in line with BSD and less with linux and plan9

I don’t know, I’d say is pretty Haiku specific :slight_smile:

Anyway, just a suggestion.

I didn’t say anything about that application, that some stuff has haiku specific implementations is hardly an argument against my assertion that we generally don’t expose system components as files, but use files as files mostly (some exceptions there, but mail and people aren’t)

The Application its Version-number was added (at last) last year or the year before. Why was the “date created” attribute was not added? Seems strange to me to add that instead of the more informative date added tag!

I imagine that it would require two dates as for a file.
Creation date would be the date of introduction of the package, the most interesting for users to display what’s new.
Then you would need a modification date, this one would indicate last time the recipe was updated and could be an indication for maintainers that a package need attention. Check for new version, check that everything is still working, etc.

I don’t think the Haiku Way is “we solve every problem with attributes and display everything in Tracker”.

Queries and filesystem attributes are a powerful tool when you deal with files in the first place. It’s not the case here. So we should rather think about a solution that does not involve running an extra daemon and then exposing the data in a less convenient way.