Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

barracuda156

macrumors 68020
Original poster
Sep 3, 2021
2,398
1,591
Has anyone been able to build i2p and Signal for PowerPC?

i2p has a version in C++ which in principle should build: https://i2pd.readthedocs.io/en/latest/
GUI version requires Qt5, so unless someone either backports it to Qt4 or fixes Qt5 for ppc, we are out of luck: https://github.com/PurpleI2P/i2pd-qt

I have tried building `i2pd` now, and it failed on this: https://github.com/PurpleI2P/i2pd/issues/1726
This is under Rosetta, I will try the same on 10.6 PPC. Sometimes it is easier than on 10.6.8.

Haven't yet dealt with building Signal, but it does not seem promising TBH: https://github.com/signalapp
 
  • Like
Reactions: lepidotós
There is some progress with i2pd finally: https://github.com/PurpleI2P/i2pd/issues/1726
It does not work yet, but at least I got the binary built:

Code:
36-227:~ svacchanda$ port -v installed i2pd
The following ports are currently installed:
  i2pd @2.41.0_0 (active) requested_variants='' platform='darwin 10' archs='ppc' date='2022-03-03T16:49:49+0800'

This is my local port, so it won’t work if you try "sudo port -v install i2pd" yet. However if we fix it and it is confirmed to build and function at least on 10.5.8 (later OS with x86 arch should be trivial matter), I will try committing it to Macports.
 
  • Like
Reactions: lepidotós
Has anyone been able to build i2p and Signal for PowerPC?

i2p has a version in C++ which in principle should build: https://i2pd.readthedocs.io/en/latest/
GUI version requires Qt5, so unless someone either backports it to Qt4 or fixes Qt5 for ppc, we are out of luck: https://github.com/PurpleI2P/i2pd-qt

I have tried building `i2pd` now, and it failed on this: https://github.com/PurpleI2P/i2pd/issues/1726
This is under Rosetta, I will try the same on 10.6 PPC. Sometimes it is easier than on 10.6.8.

Haven't yet dealt with building Signal, but it does not seem promising TBH: https://github.com/signalapp

I had a look at what macports requires for building Signal, and noting how one of its dependencies, nodejs8 (which is needed for building yarn, which is used for launching Signal), is a huge stumbling block.

It appears, at least via macports, there’s been little success of building it on any iteration of OS X/macOS:

1646541171572.png


I tried it nevertheless in Snow Leopard (10.6.8). Though the error which halts the build isn’t elucidated in the logs, my guess is it’s the need for clang-13, which needs a much later version of OS X/macOS — I’m guessing at least Mavericks, but I’m not completely certain.
 
  • Like
Reactions: barracuda156
I had a look at what macports requires for building Signal, and noting how one of its dependencies, nodejs8 (which is needed for building yarn, which is used for launching Signal), is a huge stumbling block.

It appears, at least via macports, there’s been little success of building it on any iteration of OS X/macOS:

View attachment 1968532

I tried it nevertheless in Snow Leopard (10.6.8). Though the error which halts the build isn’t elucidated in the logs, my guess is it’s the need for clang-13, which needs a much later version of OS X/macOS — I’m guessing at least Mavericks, but I’m not completely certain.

If we have PPC in mind, clang should never be used. Rather than compiler (we got gcc11 on PPC after all), nodejs indeed might be a no-go :(
 
I had a look at what macports requires for building Signal, and noting how one of its dependencies, nodejs8 (which is needed for building yarn, which is used for launching Signal), is a huge stumbling block.

It appears, at least via macports, there’s been little success of building it on any iteration of OS X/macOS:

View attachment 1968532

I tried it nevertheless in Snow Leopard (10.6.8). Though the error which halts the build isn’t elucidated in the logs, my guess is it’s the need for clang-13, which needs a much later version of OS X/macOS — I’m guessing at least Mavericks, but I’m not completely certain.

This is the closest thing to Nodejs for PPC: https://github.com/ibmruntimes/v8ppc
But it will need some work to adapt to Darwin (if at all possible, I have no idea).

P. S. As a side-note, those Macports bot builds are rather meaningless for old systems. They simply don't run those. Installation stats have more sense, but again, few people take pain of rebuilding Macports with a custom curl prefix, so essentially most of PPC builds during past year are mine LOL. That's why I was advocating actually doing that.
 
This is the closest thing to Nodejs for PPC: https://github.com/ibmruntimes/v8ppc
But it will need some work to adapt to Darwin (if at all possible, I have no idea).

P. S. As a side-note, those Macports bot builds are rather meaningless for old systems. They simply don't run those. Installation stats have more sense, but again, few people take pain of rebuilding Macports with a custom curl prefix, so essentially most of PPC builds during past year are mine LOL. That's why I was advocating actually doing that.

My point for posting that cap is more that the bot builds have been unsuccessful at building nodejs8 on any version of OS X/macOS — which, one might extrapolate, would hint at a very difficult road for trying to build the same for 10.5 or 10.6 in a PowerPC setting.

Separately, I confess/admit I want to be able to run a port install command or manually run a compile/make on a package without conducting a dissection on the code components — to do so either out-of-box or with a few, easy to understand/apply steps needed for tweaking. This may be because I’m not a computer scientist.

I applaud CS folk who can get stubborn things running in unexpected/unsupported environments, but that utility for them to be able to do so is only as meaningful as the ability for them to convey to others, in plain language, the steps needed for another person to readily repeat those conditions for their own environment. That was, in a word, what was so vexing about dealing with kencu: his reluctance and/or inability to convey that information in plain language. I mean, yes, it’s neat to know how to make something work, but: a) don’t let that capability get to your head, and b) what use is a modified version of macports for SL-PPC if only two or three people are able to get it working and not, say, twenty or thirty — much less two or three hundred — because you can’t describe how to repeat the steps in a clear, straightforward manner?

Lending from my technical writing work for clients, I can put it another way: what use is a user manual or a quick start guide if the intended audience, the end user (of even modest technical competence), can’t make heads or tails of the rough draft language the engineers used for describing a product’s (or service‘s) capabilities? Technical instructions are only as powerful as their means to communicate their steps effectively.
 
Separately, I confess/admit I want to be able to run a port install command or manually run a compile/make on a package without conducting a dissection on the code components — to do so either out-of-box or with a few, easy to understand/apply steps needed for tweaking. This may be because I’m not a computer scientist.

I applaud CS folk who can get stubborn things running in unexpected/unsupported environments, but that utility for them to be able to do so is only as meaningful as the ability for them to convey to others, in plain language, the steps needed for another person to readily repeat those conditions for their own environment. That was, in a word, what was so vexing about dealing with kencu: his reluctance and/or inability to convey that information in plain language. I mean, yes, it’s neat to know how to make something work, but: a) don’t let that capability get to your head, and b) what use is a modified version of macports for SL-PPC if only two or three people are able to get it working and not, say, twenty or thirty — much less two or three hundred — because you can’t describe how to repeat the steps in a clear, straightforward manner?

Lending from my technical writing work for clients, I can put it another way: what use is a user manual or a quick start guide if the intended audience, the end user (of even modest technical competence), can’t make heads or tails of the rough draft language the engineers used for describing a product’s (or service‘s) capabilities? Technical instructions are only as powerful as their means to communicate their steps effectively.

If you tell me which part is unclear, I will explain it in further detail step-wise.

I can assure you that no special expertise is needed at the moment to get Macports going on 10A190. Yes, it does require several manual adjustments of settings, but it takes 15 minutes and is done once.

Of course, there are ports that fail no matter what, but those also fail on 10.5.8. There are very few cases when something builds on 10.5.8 but fails on 10A190.
 
I had a look at what macports requires for building Signal, and noting how one of its dependencies, nodejs8 (which is needed for building yarn, which is used for launching Signal), is a huge stumbling block.

It appears, at least via macports, there’s been little success of building it on any iteration of OS X/macOS:

Oh, it was here, haha. I just posted in our usual thread how I built a port which was failing across the board just for the sake of making this point: https://forums.macrumors.com/thread...s.2232031/page-70?post=30966331#post-30966331
 
Update on the status on i2pd:

I got a functional binary on 10.6.8 Rosetta. Still have to fix it for 10.6 PPC, but work slowly ongoing.
 
  • Like
Reactions: eastone
I had a look at what macports requires for building Signal, and noting how one of its dependencies, nodejs8 (which is needed for building yarn, which is used for launching Signal), is a huge stumbling block.

It appears, at least via macports, there’s been little success of building it on any iteration of OS X/macOS:

View attachment 1968532

I tried it nevertheless in Snow Leopard (10.6.8). Though the error which halts the build isn’t elucidated in the logs, my guess is it’s the need for clang-13, which needs a much later version of OS X/macOS — I’m guessing at least Mavericks, but I’m not completely certain.

This is likely to be one of the big problems on Darwin PPC, but I believe it is solvable. Let’s see ;)
 
  • Like
Reactions: B S Magnet
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.