Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
What does the kernel debug kit do by the way? (Sorry if a silly question.)
It contains the debug version of the kernel and symbols etc, along with displaying more verbose information when being used to assist with debugging. It’s slower than the release versions of the kernel but is useful for testing purposes. Instructions are contained within the kit on how to replace the kernel if you want to have a go - i’ve been running the 10A190 debug kernel for a while now.
 
If anybody here has ftp access to the software repository held at Beta Archive, there are several builds available there that we are missing.

The builds listed there are:

10A096
10A190
10A222
10A261
10A286
10A314
10A354
10A380
10A403
10A432

If you are able to download and share it would be a great help to those of us working on getting Snow Leopard on PowerPC up and running.

Another thing to try is searching (or even asking people) in KDX and Hotline. Searching is a pain there, there is no search engines AFAIK, but servers aren't that many, so manual search is kinda feasible.
There are few people still using these, but many of them should be old-time hardcore Mac users, so there are chances, I guess.

P. S. If anyone already done that, please update us, so we save time and avoid double efforts.
 
I was googling in an attempt to find later builds, and this popped up: http://macosrumors.com/2009/02/10/macosx-106-snow-leopard-1st-peek/3/
Apparently 10A261 did actually run on PowerPC, at least on G5.
View attachment 1904635

I think we should read this again very carefully before anyone gets ahead of themselves:

"both developer seed 10A261 and internal builds approaching 10A270 had comparable performance and stability".

This might be a phrasing mistake that was supposed to mean "developer seed 10A261 and internal builds approaching 10A270 both had comparable performance and stability" as in on their respective platforms, i.e Mac Pro (10A261) and G5 (internal builds approaching 10A270)

Just being the devil's avocate here :)

At least we now have confirmation that there was an internal development fork that maintained ppc support for an extended period of time. And if the above doesn't turn out to be a mistake, possibly a UB build of 10A261. Has anyone found out and contacted whoever posted this?

Cheers,
 
Last edited:
I want to try building a cross-compiler for 10.6 PPC to be used on an Intel Mac.

There is a brief tutorial here: http://maniacsvault.net/articles/powerpccross (obviously without considering 10.6).
As an initial prerequisite it is needed to extract certain components from old versions of Xcode and install them into a modern Xcode on an Intel Mac. For that purpose there is this script: https://github.com/devernay/xcodelegacy

Now, that will perhaps be useless as is, since what we need is lacking in official Xcode releases, which the script uses. We do have Xcode in 10A190 distributive, but it is in a form of optional pkg installer, not standalone dmg.

Which is why my initial thought of simply replacing all references to Xcode 3.2.6 in the script text with 3.2.0 won't work.
Can anyone advise how to proceed? (Probably it may work to manually follow the script step by step and copy components via root, by I am afraid I will screw something up.)

P. S. Has anyone tried using Xcode from 10A222 or 10A261? Are they unusable on PPC? Cannot check right now, my Quad is in the office.
 
I want to try building a cross-compiler for 10.6 PPC to be used on an Intel Mac.

There is a brief tutorial here: http://maniacsvault.net/articles/powerpccross (obviously without considering 10.6).
As an initial prerequisite it is needed to extract certain components from old versions of Xcode and install them into a modern Xcode on an Intel Mac. For that purpose there is this script: https://github.com/devernay/xcodelegacy

Now, that will perhaps be useless as is, since what we need is lacking in official Xcode releases, which the script uses. We do have Xcode in 10A190 distributive, but it is in a form of optional pkg installer, not standalone dmg.

Which is why my initial thought of simply replacing all references to Xcode 3.2.6 in the script text with 3.2.0 won't work.
Can anyone advise how to proceed? (Probably it may work to manually follow the script step by step and copy components via root, by I am afraid I will screw something up.)

P. S. Has anyone tried using Xcode from 10A222 or 10A261? Are they unusable on PPC? Cannot check right now, my Quad is in the office.
I have instead been compiling and updating the toolchain as needed. The only working version of Xcode, from the seed builds, is the Beta Xcode 3.2 from 10A096 - as it is an early beta version you may wish to review the seed notes for later builds and patch/fix based on the Apple revisions. Honestly, i've had little luck actually within the Xcode IDE and have preferred to build from the command-line.

With regard to cross-compiling from X86_64, i have tried to build on 10.6 Intel before but attempting to use those binaries resulted in errors. You will need the 10.4sdk set as target if you wish to build for PowerPC on intel as the 10.4u, 10.5 and 10.6 sdks do not house all of the necessary PowerPC symbols and headers. I suspect the errors that were generated on my previous attempts were due to linking against the frameworks and libraries of the host machine and not specifying otherwise - more investigation into the correct flags is needed. I have looked at xcode legacy previously and there are a few other tutorials discoverable via google for setting-up a cross-toolchain, some geared more toward Lion/Mountain Lion as they didn't support PowerPC Compilation out-of-the-box, but i don't see why it shouldn't be possible. I'm keen to get cross-compilation working as compile time is much shorter on the MacBook than the PowerBook, iBook or G4/G5 Powermacs. I think the beta compilers are slow not just that the machines are less powerful, because compiling under 10.5.8 on the same systems is much faster. Work in progress.

I have come across a couple of interesting archives named; Developer-Tools-10A394.tar.gz and Developer-Tools-10A411.tar.gz. They each contain a password protected pdf, which i can't access, along with the source code for the tool-chain changes corresponding to the named build. Unfortunately i haven't been able to find the ones for the other builds yet, but diffing the sources to the 10.6 Open Source Projects on AOSP webpage may shine some light on what specifically was changed under the hood regarding the Developer Tools at least.

On a side note, my test update that i released previously needs an update of its own. I've discovered that the included LibXML2.dylib hasn't been compiled with flat namespace enabled and so UPDATE_DYLD_SHARED_CACHE doesn't work properly. I have recompiled with the additional LDFLAG and the error is now fixed so if you've installed the previous update i recommend installing the next one when its made available.
 
On a side note, my test update that i released previously needs an update of its own. I've discovered that the included LibXML2.dylib hasn't been compiled with flat namespace enabled and so UPDATE_DYLD_SHARED_CACHE doesn't work properly. I have recompiled with the additional LDFLAG and the error is now fixed so if you've installed the previous update i recommend installing the next one when its made available.

Yes, please, share it once it is ready.
 
  • Like
Reactions: ChrisCharman
I think we should read this again very carefully before anyone gets ahead of themselves:

"both developer seed 10A261 and internal builds approaching 10A270 had comparable performance and stability".

This might be a phrasing mistake that was supposed to mean "developer seed 10A261 and internal builds approaching 10A270 both had comparable performance and stability" as in on their respective platforms, i.e Mac Pro (10A261) and G5 (internal builds approaching 10A270)

Just being the devil's avocate here :)

At least we now have confirmation that there was an internal development fork that maintained ppc support for an extended period of time. And if the above doesn't turn out to be a mistake, possibly a UB build of 10A261. Has anyone found out and contacted whoever posted this?

Cheers,
Our main hope for any internal 'daily' of the 432 builds, that must have existed, is to find anything closed-source that remained PowerPC compatible. All of the open sources we can manually update with time and effort, but without the internal builds we will not be able to update any of the proprietary objects, such as Finder for example.

The previous articles on that website discuss PowerPC development as somewhat active but most likely not fully maintained and almost certainly full of unfixed bugs due to the focus at Apple clearly moving toward an intel and ARM release as development progressed, and if a fix for intel or ARM broke something on PowerPC it certainly wouldn't have been a high priority to remedy.

I'm personally convinced that Apple would have maintained at least a semi-working fork for PowerPC for the 'just-in-case' scenario, maybe even just for the base system, though how complete said builds might be is anybody's guess. Presumably what remained of the skeleton PowerPC team of Engineers that didn't move on to other projects would have been primarily focused on 10.5.X updates.

Apple NDA and rigorous secrecy and security measures aside, we can only hope that someone sympathetic, somewhere, is sitting on one or more of the internal builds and has somehow been made aware of this thread.

Maybe Gizmodo found one in a bar?
 
P. S. Has anyone tried using Xcode from 10A222 or 10A261? Are they unusable on PPC? Cannot check right now, my Quad is in the office.

On a lark, I just decided to install the 10A222 Xcode dev tools in a 10A96 environment to see whether a) it installed, and b) the updated Developer applications are still UB.

It may be that my already having Xcode 3.2 (the one which came with Build 10A96) prevented a proper install, as it failed at the same spot: installing the MacOSX10.4u.sdk directory. I tried twice, the second time after moving the pre-existing SDKs to the trash:

1636413217497.png


Next thing I decided to try was installing the 10A261 Xcode tools, but there was something fundamentally broken from the outset, as the Xcode installer cannot find the directory with all the Xocde/Developer installers (despite them being located in a directory at the same directory level):

1636413381286.png


Unless the Xcode installers bundled with 10A222 or 10A261 happen to be busted for everyone, then it might be an issue with my particular machine having a slightly earlier Xcode 3.2 already installed and it not liking that at all (and not simply clobbering over the previous install as if it wasn’t there in the first place).
 
I'm personally convinced that Apple would have maintained at least a semi-working fork for PowerPC for the 'just-in-case' scenario, maybe even just for the base system, though how complete said builds might be is anybody's guess. Presumably what remained of the skeleton PowerPC team of Engineers that didn't move on to other projects would have been primarily focused on 10.5.X updates.

Apple NDA and rigorous secrecy and security measures aside, we can only hope that someone sympathetic, somewhere, is sitting on one or more of the internal builds and has somehow been made aware of this thread.


Following the mentions of a PowerPC-capable build which @barracuda156 found last week, I am also in agreement with you that Apple maintained a PowerPC/UB fork of Snow Leopard as a contingency backup — in particular for the event that consumer backlash was more fierce than it ended up being; that the great recession hadn’t coincided with Snow Leopard’s development; that iPhone sales trailed off after the hype of 2007’s introduction; and/or Intel dragged their feet on subsequent, 64-bit-complete chips/EFI/logic board architectures anticipated for forthcoming Mac models succeeding the early/mid/late 2008 Macs.

I also concur with you that coaxing anyone who might have had one of those UB/PPC internal fork builds at any stage, who is bound by old NDAs, is loath to quietly and anonymously leak a .dmg of one given the spectre of an overzealous corporate legal department with too many resources at hand, even some ten-plus years after Apple ditched and abandoned Snow Leopard as the current OS X iteration. Thing is, during that time, there wouldn’t necessarily have been anything to digitally “fingerprint” the source of a leaked internal, UB build of Snow Leopard, given how any cryptographic measures to make moving a copy out of the “Loop” wouldn’t have really been a “thing” at that point (relative to, say, stuff Apple are producing nowadays). And to leak one now, so long after Apple’s abandonment, is likely pretty low on their secrecy food-chain to do that much about it.

Of course, someone who was at Apple then, during the 2007–2011 window, and involved with Snow Leopard, would know better than any of us.


Maybe Gizmodo found one in a bar?

ZING! Now that’s a vintage reference. :D
 
Following the mentions of a PowerPC-capable build which @barracuda156 found last week, I am also in agreement with you that Apple maintained a PowerPC/UB fork of Snow Leopard as a contingency backup
I know it’s offtopic but I wonder when exactly Apple started with porting full Mac OS X to ARM as a plan B. As early as 2010, when their first custom ARM design saw the light of day? If so, the port would have been of Snow Leopard.
 
  • Like
Reactions: pc297
I know it’s offtopic but I wonder when exactly Apple started with porting full Mac OS X to ARM as a plan B. As early as 2010, when their first custom ARM design saw the light of day? If so, the port would have been of Snow Leopard.
I could be wrong but I think it was an IT master's project. Off the top of my head I would say indeed around 2010. Will look it up again
 
I know it’s offtopic but I wonder when exactly Apple started with porting full Mac OS X to ARM as a plan B. As early as 2010, when their first custom ARM design saw the light of day? If so, the port would have been of Snow Leopard.

Sorry BSc thesis, but indeed 2010


And it was Darwin, albeit the version number is unknown, however this was done using SL. Interestingly, he did it using build 10A411:

"Build Built On: SnowLeopard 10A411"

for which, he writes:

"the C++ runtime had been changed for the x86_64 platform. This involved relocation several symbols to another part of the kernel. Unfortunately, the way this was implemented in the header was #if __i386__ || __ppc__" referring to issues for initial cross-compilation from what I understand.

Could this mean that cross-compilation to ppc became unsupported as of 10A411 only?
 
I know it’s offtopic but I wonder when exactly Apple started with porting full Mac OS X to ARM as a plan B. As early as 2010, when their first custom ARM design saw the light of day? If so, the port would have been of Snow Leopard.
On another note, one could consider that it started as far back the development of the iPhone, as iOS is based on Darwin for ARM
 
On another note, one could consider that it started as far back the development of the iPhone, as iOS is based on Darwin for ARM
Yes, but I don’t mean barebones Darwin or iOS. I mean full-blown Mac OS X. Like the “secret” Intel port.
 
Yes I remember reading this paper, i think i still have it in my downloads folder for reference. The port he was specifically working on was for a development board using an unsupported ARM chip - he mentions working alongside another internal ARM project which was most likely the iOS branch of the development team.

Though there are differences in his approach and goals to ours, there a probably some useful takeaways in his findings.
 
and/or Intel dragged their feet on subsequent, 64-bit-complete chips/EFI/logic board architectures anticipated for forthcoming Mac models succeeding the early/mid/late 2008 Macs.
I concur on this, especially given that until SL Intel was fully out of the bag, the only platform that had full 64-bit OSX support was ... the G5 platform ;)

So any ppc SL in the "just-in-case" drawer would have accounted for possible issues/delays in the development of the 64-bit Intel kernel and base system. They would have indeed made this a contingency plan right from the start of the development and later tossed it aside, relegating ppc development to a fork consisting of internal builds as 64-bit Intel development was making good progress.
 
Last edited:
On a lark, I just decided to install the 10A222 Xcode dev tools in a 10A96 environment to see whether a) it installed, and b) the updated Developer applications are still UB.

It may be that my already having Xcode 3.2 (the one which came with Build 10A96) prevented a proper install, as it failed at the same spot: installing the MacOSX10.4u.sdk directory. I tried twice, the second time after moving the pre-existing SDKs to the trash:

View attachment 1906151

Next thing I decided to try was installing the 10A261 Xcode tools, but there was something fundamentally broken from the outset, as the Xcode installer cannot find the directory with all the Xocde/Developer installers (despite them being located in a directory at the same directory level):

View attachment 1906156

Unless the Xcode installers bundled with 10A222 or 10A261 happen to be busted for everyone, then it might be an issue with my particular machine having a slightly earlier Xcode 3.2 already installed and it not liking that at all (and not simply clobbering over the previous install as if it wasn’t there in the first place).

I will try to do the same sometime this week, maybe even tomorrow. Since I killed my 10.5.8 install in order to try Sorbet Leo (which failed to install anyway), I can clean-install 10A190 on that part of the partition and try running later Xcode without having installed one from 10A190 DVD.
 
Next thing I decided to try was installing the 10A261 Xcode tools, but there was something fundamentally broken from the outset, as the Xcode installer cannot find the directory with all the Xocde/Developer installers (despite them being located in a directory at the same directory level):

View attachment 1906156

What happens if we run specific packages instead of trying to update the whole thing? Or it just is not designed to work this way in any case?
 
"Build Built On: SnowLeopard 10A411"

for which, he writes:

"the C++ runtime had been changed for the x86_64 platform. This involved relocation several symbols to another part of the kernel. Unfortunately, the way this was implemented in the header was #if __i386__ || __ppc__" referring to issues for initial cross-compilation from what I understand.

Could this mean that cross-compilation to ppc became unsupported as of 10A411 only?

I doubt it. Owing to what we are beginning to understand was probably a diligent effort internally to have contingencies ready, it appears more likely that Apple, at least through the 10A432 “Golden Master”, maintained some kind of multi-architecture production stream which, at a minimum, also included a UB version with support for PowerPC Macs. I have no idea whether Apple also devoted human resources to maintaining a skeleton build for ARM, but I suppose if they wanted to, they would have had the knowledge and resources to probably do so.

I suspect, however, that even a UB fork, all the way to 10A432, probably came up short for functional components like hardware graphics support for AGP video cards on pre-PCIe G5s in the PowerPC line of Macs, and OpenCL support on non-Intel architecture was probably fairly limited. I mean, it’s entirely possible they turned to a longtime internal specialist for GPU-related stuff who could have been side-tasked with creating a kludge or legacy shim hardware graphics support for that UB fork, but at this point there is nothing known to support that hypothesis.

I also have no insight, just like y’all, on whether Apple maintained internal support for a UB Snow Leopard beyond 10A432 — meaning, once the DVD for retail was approved for replication, all subsequent work on an internal UB build might have come to a close. Consequently, minor version updates would probably have been limited to Intel architecture only. But again, take all of this as only conjecture at best and fiction at worst.

EDIT to add: After reviewing that thesis, it appears the publicly (i.e., ADC) available 10A411 seed update build was used.

I concur on this, especially given that until SL Intel was fully out of the bag, the only platform that had full 64-bit OSX support was ... the G5 platform ;)

Yes and no.

From examining binaries in the SL betas, we have seen an abundance of either “dual-architecture” (i386 and x86_64) or “tri-architecture” (i386, x86_64, and ppc) builds, but very few, if any, which also call out “ppc64” architecture. Separately, considering how Snow Leopard was still being built for 32-bit architecture despite being re-engineered for 64-bit complete support on systems with all-64-bit architecture and firmware (i.e., 64-bit Intel), an internal fork with UB support for PowerPCs probably kept that arm to 32-bit support to reach the broadest variety of PowerPC Macs in circulation.

Put another way: there was zero financial or strategic incentive for them to bring in a fourth architecture for Snow Leopard, even as a contingency, and making its binaries “quad-architecture” (i386, x86_64, ppc, and ppc64). If the UB contingency had to be used, Apple could still market Snow Leopard as a 64-bit OS from the ground up because, yes, it was 64-bit from the ground up for Intel 64-bit Macs. The asterisk in the fine print would then clarify that it did not apply to 64-bit PowerPC Macs such as all the G5s.


So any ppc SL in the "just-in-case" drawer would have accounted for possible issues/delays in the development of the 64-bit Intel kernel and base system. They would have indeed made this a contingency plan right from the start of the development and later tossed it aside, relegating ppc development to a fork consisting of internal builds as 64-bit Intel development was making good progress.

Correct.

What happens if we run specific packages instead of trying to update the whole thing? Or it just is not designed to work this way in any case?

I tried this with the Xcode installer for 10A261, and it griped that the .pkg installer (I forget which one I tried) was not intended to be applied in that manner. Part of the Xcode installer contains a script which predetermines which secondary packages are to be used and which are to be skipped.
 
Last edited:
  • Like
Reactions: pc297
I have no idea whether Apple also devoted human resources to maintaining a skeleton build for ARM, but I suppose if they wanted to, they would have had the knowledge and resources to probably do so.

[...]

EDIT to add: After reviewing that thesis, it appears the publicly (i.e., ADC) available 10A411 seed update build was used.

[...]

Separately, considering how Snow Leopard was still being built for 32-bit architecture despite being re-engineered for 64-bit complete support on systems with all-64-bit architecture and firmware (i.e., 64-bit Intel), an internal fork with UB support for PowerPCs probably kept that arm to 32-bit support to reach the broadest variety of PowerPC Macs in circulation.


Indeed, one could wonder why he would use 10A411 over 10A432 when the latter had already been out for a year? The MV88F6281 was 32-bit ARM, maybe he used a build that still had 32-bit ppc headers etc to cross-compile to 32-bit ARM? If what @WindowsXPuser says about 10A428 (or likely an equivalent build from a derived UB fork) is true, in that it was the last one to support 32-bit PPC, that would make sense. All handwaiving though, obviously. I will try to see if it's out there and inspect the binaries to see how much ppc code is left in it, in case it is available (albeit apparently not according to the wiki).

EDIT: as @B S Magnet noted, this was a seed update from previous WWDC builds, so unlikely to have any ppc code left in it


From examining binaries in the SL betas, we have seen an abundance of either “dual-architecture” (i386 and x86_64) or “tri-architecture” (i386, x86_64, and ppc) builds, but very few, if any, which also call out “ppc64” architecture.

find . -type f -perm +111 -exec lipo -info {} \; | grep ppc64 | wc -l

reveals 24 hits on 10A190. On the other hand, on my 10.5.8 install, the same command reveals 1140 hits, although it's not really a clean install anymore. Indeed doesn't sound like ppc64 was sitting on top of their to-do list for SL.
 
Last edited:
  • Like
Reactions: ChrisCharman
Indeed, one could wonder why he would use 10A411 over 10A432 when the latter had already been out for a year?

It appears the author’s time at 2 Infinite Loop occurred in 2009, between 6 July and 25 September (Plan of Approach, page 32 in the PDF).

This would place the timing for getting acquainted with internals and other NDA-bound stuff out of the way during the first week, and access to the 10A411 development seed ultimately released to ADC membership on or about 15 July. This was probably the build which was ready for visiting third parties to be able to tinker within.

If you’ve not completed a thesis before, there’s often a significant gap between when the field research/data collection happens and when the analysis and synthesizing of those findings go into a final written report, or thesis, approved by one’s advisor(s)/supervisor(s). For something like this, the proposal for this idea probably germinated sometime during the 2008–09 academic year and the internship (when the field research occurred), being mid-2009, preceded the drafting, revisions, and accepted submission of the final report as part of earning a master accreditation in compsci (I’m guessing) at the end of the 2009–10 academic year.

The MV88F6281 was 32-bit ARM, maybe he used a build that still had 32-bit ppc headers etc to cross-compile to 32-bit ARM?

According to page 8, the build being used for this research appeared to not only come with a “vanilla ARM toolchain”, but also the ARMv5 compiler (despite this 10A411 build also having libraries for ARMv6 and ARMv7). An ARMv5 compiler would, I imagine, hint at ARM-specific headers, probably (or possibly) related to ongoing iOS development internally. Whether the flavour of 10A411 used for this research had 32-bit PPC headers in it is an unknown.

If what @WindowsXPuser says about 10A428 (or likely an equivalent build from a derived UB fork) is true, in that it was the last one to support 32-bit PPC, that would make sense. All handwaiving though, obviously. I will try to see if it's out there and inspect the binaries to see how much ppc code is left in it, in case it is available (albeit apparently not according to the wiki).

Go for it. Also, I find it curious that only two days separated a “final” 32/64-bit PPC UB build and a “final” 64-bit build. I would imagine it might take more than two calendar days to strip out 32-bit PPC code. But more key: what function would doing this actually serve so late in product development?

find . -type f -perm +111 -exec lipo -info {} \; | grep ppc64 | wc -l

This gave me a lot of noise before I cancelled it:

Code:
sh-4.3# find . -type f -perm +111 -exec lipo -info {} \; | grep ppc64 | wc -l
lipo: can't figure out the architecture type of: ./Library/Automator/Add Grid to PDF Documents.action/Contents/Resources/graphpaper.py
lipo: can't figure out the architecture type of: ./Library/Automator/Combine PDF Pages.action/Contents/Resources/join.py
lipo: can't figure out the architecture type of: ./Library/Automator/Extract Odd & Even Pages.action/Contents/Resources/extract.py
lipo: can't figure out the architecture type of: ./Library/Automator/Get Image URLs from Webpage.action/Contents/Resources/images
lipo: can't figure out the architecture type of: ./Library/Automator/Get Link URLs from Webpages.action/Contents/Resources/links

Indeed doesn't sound like ppc64 was sitting at the top of their pile of their list of things to do for SL.

Not with the build fork going out into the world via ADC/WWDC, no.
 
  • Like
Reactions: ChrisCharman
This gave me a lot of noise before I cancelled it:

Code:
sh-4.3# find . -type f -perm +111 -exec lipo -info {} \; | grep ppc64 | wc -l
lipo: can't figure out the architecture type of: ./Library/Automator/Add Grid to PDF Documents.action/Contents/Resources/graphpaper.py
lipo: can't figure out the architecture type of: ./Library/Automator/Combine PDF Pages.action/Contents/Resources/join.py
lipo: can't figure out the architecture type of: ./Library/Automator/Extract Odd & Even Pages.action/Contents/Resources/extract.py
lipo: can't figure out the architecture type of: ./Library/Automator/Get Image URLs from Webpage.action/Contents/Resources/images
lipo: can't figure out the architecture type of: ./Library/Automator/Get Link URLs from Webpages.action/Contents/Resources/links

It does eventually complete and show the number of hits despite the warning messages about files that it can't identify, takes 5-10 mins to run

Cheers,
 
A quick update on later version of Xcode. Good news, Xcode from 10A261 installed on 10A190 (clean system with no Xcode). Bad news, Xcode app crashes on launch, though it is seen as being executable and attempts to start.

GCC is newer than from Xcode from 10A190.
 

Attachments

  • Screenshot 2021-11-10 18-40-01.png
    Screenshot 2021-11-10 18-40-01.png
    167.3 KB · Views: 76
A quick update on later version of Xcode. Good news, Xcode from 10A261 installed on 10A190 (clean system with no Xcode). Bad news, Xcode app crashes on launch, though it is seen as being executable and attempts to start.

GCC is newer than from Xcode from 10A190.
Is the GCC PowerPC compatible? Perhaps the parts of the 10A261 Xcode that are failing can be patched with the working components from 10A096 version, or recompiled?
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.