Making it more sustainable to keep software repos up to date?

[randomshinichi]

I originally posted this idea on https://github.com/mistydemeo/tigerbrew/issues/719 but mistydemeo seems to be too busy to check, or maybe he's not interested, I dunno. So I'm asking the community here if there's interest in such an effort.

Basically, it's obvious that Tigerbrew is falling behind.
(doesn't include everything, nor does it even have bottles for the Formula that work. I was very glad to see that OpenSSH 8.4 compiles on Leopard even though there was no formula for it, and IIRC port refused to compile it because of some missing libxml2 dependency or something. MacPorts doesn't care about 10.5 PPC anymore so I had to ignore it.)

If mistydemeo doesn't have time for it, then the Tigerbrew infrastructure should be changed to make contributing to the project effortless, even incentivized. Enough people are still interested in the PPC platform for this to be possible, I think.

Proposal: When Tigerbrew is first setup, it prompts for the user's "nickname", and whether to upload bottles if a Formula finishes compiling. If the user agrees, bottles are uploaded to a temporary repository. If the sha256sum already exists, it is not uploaded. A simple microservice can be written here, and I can provide some storage.


The nickname is used to give credit for the upload. Let's say 1 point for each bottle uploaded. Sounds silly but it works: look at Folding@Home points, which are worthless but people still buy graphics cards to top the charts.


Thus the problem of compiling binaries for different architectures and OSX versions is spread across the community automatically.


The temporary repository is used for some semblance of control over the binaries (in case someone tries to upload malware or whatever). IPFS can solve the problem of hosting. People with little technical knowhow can be told they can still contribute to this project simply by running a IPFS node and gateway that hosts the bottles. Periodically we will need to delete old bottles. IPFS nodes can independently choose to keep the old bottles or follow the crowd and save disk space.


I'd be happy to help. I don't want to compile stuff on my iBook anymore. I just want to use it.

 

[Project Alice]

I originally posted this idea on https://github.com/mistydemeo/tigerbrew/issues/719 but mistydemeo seems to be too busy to check, or maybe he's not interested, I dunno. So I'm asking the community here if there's interest in such an effort.

Basically, it's obvious that Tigerbrew doesn't include everything, nor does it even have bottles for the Formula that work. I was very glad to see that OpenSSH 8.4 compiles on Leopard even though there was no formula for it, and IIRC port refused to compile it because of some missing libxml2 dependency or something. MacPorts doesn't care about 10.5 PPC anymore so I had to ignore it.

If mistydemeo doesn't have time for it, then the Tigerbrew infrastructure should be changed to make contributing to the project effortless, even incentivized. Enough people are still interested in the PPC platform for this to be possible, I think.

I can't even get TigerBrew to install. It fails on some SSL error every time.
It's installed on my PowerMac G5, though. I don't remember installing it but obviously I did at one point. I can't figure out how to get it working on anything else though.

MacPorts fails to build a lot of things. libiconv for example fails to build on every machine I've tried it on (except one, my 17" PB, no idea why but it fails on everything else). The same ports that tend to fail on macports, works on brew thats installed on the one G5.

 

[randomshinichi]

I can't even get TigerBrew to install. It fails on some SSL error every time.
It's installed on my PowerMac G5, though. I don't remember installing it but obviously I did at one point. I can't figure out how to get it working on anything else though.

MacPorts fails to build a lot of things. libiconv for example fails to build on every machine I've tried it on (except one, my 17" PB, no idea why but it fails on everything else). The same ports that tend to fail on macports, works on brew thats installed on the one G5.

I'm thinking of making a closed economy. You buy some tokens to be able to use a particular software repo (say a fork of Tigerbrew). In return the software is up to date as upstream allows, and you can find almost everything there (that upstream issues allow). And if enough people participate in this economy, maybe this community can sponsor powerpc darwin support in other software projects.

Would getting tokens (albeit with some minimal monetary value) actually motivate you to spend some free time solving issues like these? e.g. compiling bottles, fixing Tigerbrew formulae, let's try to port that to Tiger/Leopard...

This has been done before with Folding@Home and their points system. It certainly works even though there is no monetary value behind those points - although this is because the computer does the work, all you have to do is install it and make sure it's running. In this case, more human interaction is needed, so it's more important to build a proper economy, even if it is closed. (closed meaning not exchangeable for monetary value)

 

[Wowfunhappy]

MacPorts fails to build a lot of things. libiconv for example fails to build on every machine I've tried it on (except one, my 17" PB, no idea why but it fails on everything else). The same ports that tend to fail on macports, works on brew thats installed on the one G5.

Have you tried opening trac tickets for these? I've found the MacPorts devs to be very responsive—just make sure to included the needed information and logs and such so that the maintainers will have the information they need.