Oolite Bulletins

Cholmondely wrote: ↑Sat Apr 15, 2023 11:59 pm

hiran wrote: ↑Sat Apr 15, 2023 8:21 pm

...@Cholmondely, I will need your support. You have access to Mac devices and experience of building software there.

Personally speaking, I will be delighted to give you whatever help I can...

I'm not too sure about my supposed "experience of building of software" though! But I'm sure that Commander X has it in spades.

We'd have to start off in quite deep s***. If we try to build Oolite the way it used to be built it will all fail.
Then we have to see where compilation problems arise, and probably merge Ahruman's PR - which might still fail to build Oolite.
Then we still have to see where compilation problems arise and start modifying the code (with replacing old libraries where necessary) before we can get a build that is worth testing.

During that exercise we need to modify lots of code. It should be better if we introduce automated unit testing before touching too much - just as a safeguard.

The last attempts I made happened in a VirtualBox VM I have around. Since then, though, I migrated most of my VMs to QEMU. The migration was not an actual conversion, but mainly a re-installation of each OS. As of now, I don't have a dev environment for Mac OS X, but that can be rectified.

From what I recall, the main issue seems to be with the UI (the .xib file). The existing file has an old format, and I think that some of the information the older format dealt with is lost during import/conversion to the newer format, which leads to incorrect initialization of OpenGL.
In order to build Oolite with newer versions of Apple's toolchain, someone with more experience and dedication in development on Mac OS X/macOS should look into how the conversion should happen or just re-build the UI with the new tools.

Commander_X wrote: ↑Mon Apr 17, 2023 4:15 pm

In order to build Oolite with newer versions of Apple's toolchain, someone with more experience and dedication in development on Mac OS X/macOS should look into how the conversion should happen or just re-build the UI with the new tools.

Thank you for the feedback.
Well, unless this someone shows up we will not be able to bring back Mac builds.

Should this be the time we accept that fact and officially drop Mac support?

hiran wrote: ↑Mon Apr 17, 2023 7:16 pm

Commander_X wrote: ↑Mon Apr 17, 2023 4:15 pm

In order to build Oolite with newer versions of Apple's toolchain, someone with more experience and dedication in development on Mac OS X/macOS should look into how the conversion should happen or just re-build the UI with the new tools.

Thank you for the feedback.
Well, unless this someone shows up we will not be able to bring back Mac builds.

Should this be the time we accept that fact and officially drop Mac support?

Neither do we move forward in this thread nor in the GitHub issue https://github.com/OoliteProject/oolite/issues/360.
I suggest we officially drop Mac support. If people scream and get active that decision can still be reverted.

Say it ain't so, Joe! I can take a look at what it would take. Give me some time to review the issues and see about building on my various Macs. I have some (now about 13 years old!) iOS development experience and a variety of various (mostly older x86, but one Apple M2 2023 MacBook Pro) macs around the house.

Bogatyr wrote: ↑Fri Sep 13, 2024 9:21 am

Say it ain't so, Joe! I can take a look at what it would take. Give me some time to review the issues and see about building on my various Macs. I have some (now about 13 years old!) iOS development experience and a variety of various (mostly older x86, but one Apple M2 2023 MacBook Pro) macs around the house.

At our get-together (7th July in London), Aegidian was reminiscing on how supportive Apple used to be - compared to the current situation with missing or poor documentation.

The last published versions for the AppleMac were engendered by Getafix (1.90 & an old 1.91 - he did the Windows/Linux, too).

Phkb recently (this year) had a couple of failed stabs at it.

OK so the situation really seems pretty dire. Oof.

Apple is (was) great when it works, and pretty horrible when it doesn't. Wind-blows generally sucks all around but there's usually some way to get what you want to work, to work.

So there are no cases of any successful builds of even Intel Mac in recent history?

Bogatyr wrote: ↑Fri Sep 13, 2024 11:39 am

So there are no cases of any successful builds of even Intel Mac in recent history?

phkb managed to successfully compile the current 1.91 master but the binaries don't seem to run correctly and fail on startup with OpenGL related errors. See discussion here.

Thanks, I'll review the thread.

So admittedly I'm only here once every couple years when I've been long enough away from Oolite to get the urge to build up an iron ass from nothing again, but it's always been a hallmark of Oolite that it ran on whatever platform was my main one at the moment -- Mac, or Windows, the great majority of the time it's been Mac, and it ran on whatever I had at the moment. I'm sure there was a discussion about the renderer upgrade to OpenGL 3.3, but does that mean that old hardware is automatically out of the picture?

I guess I'm going to be stuck on Oolite 1.90 then, since Macs are the staple platform still in my house...

One thing to think about, is that Apple's decision to make OpenGL problematic, may be the but harbinger of things to come, perhaps in the near future, and that there may be no safe harbor on any platform or GPU, and that looking to an API independent renderer that can be targeted to different back-ends may be the only way forward....

Bogatyr wrote: ↑Sat Sep 14, 2024 4:49 am

I'm sure there was a discussion about the renderer upgrade to OpenGL 3.3, but does that mean that old hardware is automatically out of the picture?

No, not at all. You only need at least one developer with enough time and dedication to adapt the renderer for the older platform of their choice OGL 3.3 is supported on really old hardware at this stage.

One thing to think about, is that Apple's decision to make OpenGL problematic, may be the but harbinger of things to come, perhaps in the near future, and that there may be no safe harbor on any platform or GPU, and that looking to an API independent renderer that can be targeted to different back-ends may be the only way forward....

Porting such a thing to Oolite will be exponentially more and harder work than attempting to switch the Mac port to use OpenGL 3.3. I would even go as far as calling it next to impossible.