The rate of progress is staggering @educovas! I can’t wait to test this and read through what steps you’ve taken to achieve this. Thank you for all of the work you’ve been putting into this.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Snow Leopard on Unsupported PowerPC Macs
- Thread starter Larsvonhier
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
If there’s one fly in this ointment, I am disappointed these steps
EDIT to add: I would be delighted for someone here — anyone — to spell out, in plain language, how and why my hot take above seems controversial or unpopular.
I’ll add I’ve never been warm to the notion of taking an open community project and turning aspects of its development opaque, proprietary, and/or vague, in the name of rushed progress by one person. The approach makes it tougher for peers to test on their own hardware environments, such as G5s, by following the same steps to arrive to those milestones (or to find there are variations needed in order for wider compatibility).
This issue also includes the problem of not detailing alterations from build to build, to make them boot only on selected PowerPC Macs. In the end, I find the opaque approach is not terribly collegial or community-focussed in the long run.
But please, go off on my kvetching.
Last edited:
This might be so unstable that this bug might not even be a problem lol. And there are so many files replaced that maybe the bug doesn't even exist on this.
it's mostly the same steps I had to do to get 10A222 but with much more files replaced to make the old apps and the old Window Server work (definitely not as stable as it used to be). And I definitely can't document what I replaced because I don't even remember anything anymore, the only thing I'll be able to offer is the zip with the replaced files.
I also figured out that Ethernet has the same problem as WiFi. If I can't get an internet connection on this build, i'll probably stop at 10.6.1 since this will have no use for me without that.
I'll keep trying to fix for a couple days (maybe I'll get some help, I don't know), but if that can't be fixed, I'll share the image and replaced files without internet connection.
I'll keep trying to fix for a couple days (maybe I'll get some help, I don't know), but if that can't be fixed, I'll share the image and replaced files without internet connection.
Just in case, please upload then all versions (I mean, 10.6, 10.6.1, 10.6.2?, as you made them). Because results on different hardware may vary, so it is worth having options to try from.
I’m calling it… G4 only. G5’s will likely kernel panic given results with the D3 kernel.
1. Still worth trying, who knows.
2. It may be possible to fix it for G5, but we do not know which version will be more compatible without trying.
10.6 and 10.6.1 are 99% the same, it's a very small update that I only got like 2/3 of it because of some intel only binaries and bluetooth binaries that doesn't work with old hardware. The important stuff I was able to get was the updated libSystem, frameworks and IOSCSIArchitectureModelFamily kext. Apple didn't update the kernel so compatibility should be the same.
I’d be happy to download the zipped files and could probably reverse engineer what you’ve done and do a write up, now that you have it up and running.
If a .dmg is made available i’m sure a few people on here would be more than happy to tinker and assist in improving functionality, as has been done up until now mate. Nobody is expecting you to singlehandedly deliver something all the way to 10.6.8 bug free. We’re used to messing around with developer previews. What you have contributed thus far is greatly appreciated.
So LibSystem has been updated? That is very useful for compatibility with newer software! Also may be the cause of some of the instability given system calls and APIs changed throughout the development.
Again. Cannot overstate how excited I am to tinker with this. Thank you!
Last edited:
It’s definitely possible, now that we know the kernel is PowerPC compatible and can be cross-compiled. Will just take time and effort for those interested.
Current libSystem version is 124.1.1 so it might be enough to fix some compatibility issues. The only big issue I had with the new version was that spindump was crashing due to a missing symbol in libSystem but since it's not essential, I've just disabled it.
If there’s one fly in this ointment, I am disappointed these stepscannotwill not be documented for the wikipost. This won’t help for future open documentation reference, as there will no doubt be others in the future to wonder how we got to where we did.
EDIT to add: I would be delighted for someone here — anyone — to spell out, in plain language, how and why my hot take above seems controversial or unpopular.
I’ll add I’ve never been warm to the notion of taking an open community project and turning aspects of its development opaque, proprietary, and/or vague, in the name of rushed progress by one person. The approach makes it tougher for peers to test on their own hardware environments, such as G5s, by following the same steps to arrive to those milestones (or to find there are variations needed in order for wider compatibility).
This issue also includes the problem of not detailing alterations from build to build, to make them boot only on selected PowerPC Macs. In the end, I find the opaque approach is not terribly collegial or community-focussed in the long run.
But please, go off on my kvetching.
No one is arguing that having documentation is something undesirable, I believe. But a) given limited resources, more of A means less of B, and b) given that this is a non-commercial project based on willingness to contribute, any given individual is free to decide in regards of his/her contribution, whether A or B is to be preferred and in which proportion. (And obviously others can decide whether they are interested to use such a contribution or not.)
All this work takes ridiculous amount of time. Unfortunately, we do not have a team of engineers able and willing to work on this full-time. It would be better to have such a team. But it is not there
So having a perfect product with comprehensive documentation which gonna work out-of-the-box on every hardware is just not among feasible options.
OK, point taken.
Then, as one who has volunteered, along with doing the testing, considerable time to keeping the project transparent to anybody who wants to get involved, I ask, in return, that at least a minimum effort goes into documenting the work to make a new build (or a feature within a build) possible. Without this, it prevents additional folks interested in getting involved from doing so when they don’t have a reference guide or roadmap of where to pick up on things they might want to work on.
To re-frame this: how likely would educovas, yourself, and others had gotten involved with the SL-PPC project, as y’all have, had there not been, at minimum, a wikipost collecting together all the work done prior to getting involved — helping to illustrate what had been done and what was still not yet tried? Without that documentation, how involved or likely would y’all have spring-boarded into the project?
I know. Intimately. As do you.
And I also know it helps to be able to have the same notes amongst folks to be able to load-share/load-balance that work!
It would be better to have such a team. But it is not there
As I said to z970 long ago, this isn’t a product, but an ongoing community project — a marathon, not a sprint.
As with other community projects, such as throughout the open-source community, thorough documentation from contributors is foundational. I can’t really overstate this. And as one who’s heavily involved with macports, it’s something I know you‘re able to appreciate.
Throughout SL-PPC, I have volunteered to collect together new information as people come up with solutions and even breakthroughs (i.e., via the upkeep of the wikipost).
But in order to do that, one must have that information available in the first place.
If volunteers aren’t, you know, volunteering that documentation of their donated work, then all the effort we undertake individually, voluntarily, ends up being isolated, siloed, and largely for naught.
Without that documentation, multiple volunteers may think, unintentionally, they’re working on something which hasn’t been worked on previously, only to find out hours of work later how, yes, it had been worked on (but whose previous efforts were either poorly documented or not documented). This doesn’t provision a terribly efficient use of limited, invaluable volunteer resources.
I’ve regarded — and continue to regard — folks like you, ChrisCharman, educovas, Larsvonhier, pc597, swamprock, and many more to be an ad-hoc, informal, rolling SL-PPC team, as we’re able to give spare time in our lives toward it (hence the marathon or running relay metaphor).
And perhaps that’s also where we differ in our vantage around how we approach the transparency of what we contribute to this project, rather than to focus only on the end result, leaving everything preceding a kind of black box for everyone else. I’m trying to think about tomorrow as much as I’m thinking about right now.
I hope that resolves what I’m getting at here. Cheers.
The sad thing is that people who only have the G5’s are SOL pretty much if I had to guess. I’m going to make an educated guess based on the 10.0.0 D3 kernel that GM/10.6.1 will just kernel panic as well on said systems.
There is this handy tool
Comparing Folders
Sure, it takes some time for that app to load and calculate the differences, but then, I don't think the whole images need to be compared. Maybe only some select directories.
The problem, at least for me, is that I can not decipher which @educovas images to download and where to find them.
That Araxis thing has a 30 day trial, btw.
The only thing I can think of is to compile the kernel… but I don’t know where to begin.
I've been sharing a zip with all the replaced files and will also do that for 10.6/10.6.1, so that would not be necessary.There is this handy tool
that I've been using (albeit much older version) to spot the differences between various versions of apps. Maybe some of us, who have spare time, could check for differences in images uploaded by @educovas and write down the changes?Comparing Folders
www.araxis.com
Sure, it takes some time for that app to load and calculate the differences, but then, I don't think the whole images need to be compared. Maybe only some select directories.
The problem, at least for me, is that I can not decipher which @educovas images to download and where to find them.
That Araxis thing has a 30 day trial, btw.
I've tried everything I could to fix WiFi/Ethernet but no success so I'm giving up and I'll probably share what I have tomorrow.
I don't think I'll try to boot newer builds since it would be a waste of time for me to do all that work and not have an internet connection.
I don't think I'll try to boot newer builds since it would be a waste of time for me to do all that work and not have an internet connection.
While you're "in the groove", maybe it's worth continuing, even without the WiFi/Ethernet? Those could be revisited later.
I don't think it's worth my time to do all that work and not be able to use it because there is no internet.
If you are indeed calling it a day, please do share what you have so the rest of us can continue with what you’ve been doing. It will save a lot of time compared to trying to figure it out ourselves.
I’m sure there is a fix and it will be discovered eventually. If you do have any spare time once you stop your work on this, it would be immensely useful if you could share any information, resources, guides or tips that have aided you thus far.
Kernel can be rebuilt, and it should feasible to figure out what Apple has broken, after all.
Sorted out fetching ports sources, but very timely did a quick rsyncing before going home which apparently wiped out packages from the server LOL
(Be careful with ending slashes.)
No big deal, but uploading 45 GB will take another day.
But I guess at least by Sunday we have a way to get gcc13, latest CMake and what’s not in half an hour on a G4.