Just so nobody gets confused: This is an official Aminet statement.
Read and search INDEX [...] Download and search full .readme files for potential matches.
INDEX is close to 7 MB right now, not sure what kind of target audience you have in mind? Downloading a 7 MB file every time you run the application won't be fun on an Amiga. Not to mention the amount of memory searching this file would require.
People downloading 81000 readmes because they want to search something? That's a no-go.
Plus, you'd have to implement a mechanism to make sure the local copy stays up to date. A quick look into our changelog will show you that we're constantly cleaning up the archive and/or the directory tree, or adding new files which won't show up in RECENT.
A much more sensible approach would be to actually ask us for ideas. Here's one: We expand the existing search engine so it displays results in some machine readable format (TSV?, JSON?) and external clients use that instead: they simply download aminet.net/json/some_search_string via HTTP and parse it.
Providing a downloadable version of the entire database would be another option, of course - but not for an Amiga based solution, obviously.
I was planning a new Aminet client anyway (and searching a coder for that job), but a graphical one. For CLI access, there's ADT (Aminet Download Tool) - a console FTP client with some extras (display recent, display directory listings in Aminet format, search for filenames and descriptions).
Most if not all GUI FTP clients on the Amiga cloned these features aswell, so IMHO instead of reinventing the wheel, a better idea might be to update ADT (it doesn't compile on Unix systems anymore, IIRC) and get the search daemon on Aminet (written in Python and C) running again - that way, all Amiga FTP clients would benefit.
But tbh, I don't see the need for a console based client anymore.
Unpack archive (requires lots of extras to be able to do this on the Amiga. LhA, LZX, UnRar, UnZip, XPK, etc)
No, that actually only requires the XAD library system. Aminet doesn't support RAR or LZX anyway.
I haven't decided where, but it should also be able to eventually read a dependencies file from someplace that will reach out and grab dependencies for a given archive.
This already exists, there's a 68k port of Grunch - and it's completely dead, because maintaining the database is both a nightmare and a lot of work. The MorphOS database used to be (still is?) quite up to date, but there's simply not enough interest on 68k anymore.
Here's what I had in mind, the target audience being Amigas with moderate to low specs and some form of network access:
- graphical client, running on its own screen, mimicking the web interface
- features: Recent, directory browsing, view readmes, simple/advanced search, download one file at a time
- communicate with aminet.net using HTTP connections
- use minimal resources (RAM, Chipset, GUI elements) to make sure you're compatible with as many setups as possible
- try to stay compatible with 1.x (even if there's no TCP stack for 1.x yet - that would be step 2).
- free and open source software
Basically: get rid of the Browser (+MUI/Classact) requirement, do not rely on other computers in the household - but still offload all the resource heavy tasks to an external machine (i.e. the Aminet server).