Verrry nice!
first a random note:
So the first error(image 2) about being unable to write to the Journal can be fixed by turning journaling off in the installation(quite easy actually)...I suppose this'll incur some cost, but nbd for now.
Ok, now to the beef(or lettuce if you vegetarian):
This output shows us that kernel gets very far along before actually complaining that it can't do things to a read-only filesystem...
This may be a good place to implement a linux-like strategy of overlaying tmpfs and unionfs(which os x actually has for dmgs!) or some other remounting scheme to work around these errors...We're deep into userland at this point--Login window is actually started!
I'm going to ponder a solid solution and get back tomorrow(need to sleep). ideas?
135 replies to this topic
#41
Posted 30 October 2008 - 06:04 AM
#42
Posted 30 October 2008 - 07:03 PM
looks good. cool that it is booting past login winodw start.
Hagar - that was a usb stick right?
Hagar - that was a usb stick right?
#43
Posted 31 October 2008 - 12:59 AM
What if we have some script that starts pretty late into the boot(but before kernel tries to write anything important to disk) that creates/loads a ramdisk image and then loopback & union mounts that over / (or something else if that doesn't work). That way when we try to write things like the pids and states and other random bits, they'll be written safely to the ramdisk image!
#44
Posted 31 October 2008 - 02:12 AM
I would just make users go through a simple initialization program like some Linux distro's do.
1. Boot
--The kernel will load a script that tells which drivers to load
2. Load LoginWindow then prompt our application
--Our Application launches and tries to detect the hardware and asks us to confirm and correct.
--Our Application writes the settings to that script/file and reboots the computer
3. User boots again, now into OS X, and has an icon on the desktop that says "install to disk" just like the *buntu's
~MoC
1. Boot
--The kernel will load a script that tells which drivers to load
2. Load LoginWindow then prompt our application
--Our Application launches and tries to detect the hardware and asks us to confirm and correct.
--Our Application writes the settings to that script/file and reboots the computer
3. User boots again, now into OS X, and has an icon on the desktop that says "install to disk" just like the *buntu's
~MoC
#45
Posted 31 October 2008 - 02:34 AM
spartango: sounds good, but how would the system know to write to that mounted disk instead of the mounted dmg?
MoC: That sounds like a good idea, but we still need to get the ram disk part worked out..
A question for everyone: Would everyone want some type of wiki that we could work on? I could set up one if we wanted...
Edit: Found this on creating a ramdisk:
http://osxdaily.com/...sk-in-mac-os-x/
MoC: That sounds like a good idea, but we still need to get the ram disk part worked out..
A question for everyone: Would everyone want some type of wiki that we could work on? I could set up one if we wanted...
Edit: Found this on creating a ramdisk:
hdid -nomount ram://52428800 newfs_hfs /dev/disk1 mkdir /tmp/ramdisk1 mount -t hfs /dev/disk1 /tmp/ramdisk1
http://osxdaily.com/...sk-in-mac-os-x/
#46
Posted 31 October 2008 - 02:42 AM
either a loopback mount, symlinks, or unionfs
#47
Posted 31 October 2008 - 02:50 AM
Ok. That sounds really feasible. What can I do to help?
Any thoughts on the wiki idea?
Any thoughts on the wiki idea?
#48
Posted 31 October 2008 - 03:18 AM
Hello
Not to rush things, but when can we expect a release?
Panarchy
Not to rush things, but when can we expect a release?
Panarchy
#49
Posted 31 October 2008 - 03:23 AM
We wish we knew. You'll be updated of our progress on the homepage (by REVENGE I assume)
#50
Posted 31 October 2008 - 05:24 AM
#51
Posted 31 October 2008 - 05:25 AM
Umm... I'm a try to formulate this idea starting sometime this weekend, as time permits. Then will test and share and test more.
This definitely a good start, and a wiki'd be useful soon too, tho not yet.
@Panarchy: we started the new team like a week ago...and we have good ideas but no results yet.
This definitely a good start, and a wiki'd be useful soon too, tho not yet.
@Panarchy: we started the new team like a week ago...and we have good ideas but no results yet.
#52
Posted 31 October 2008 - 05:53 AM
Yeah. I think we have a great start, with some great possiblilities but it is going to take time.
I will be able to setup a wiki for us when we want.
I will be able to setup a wiki for us when we want.
#53
Posted 31 October 2008 - 11:51 AM
Well we have that worked out, let's split the labour and say who does what?
#54
Posted 31 October 2008 - 11:56 AM
Well, people wanting release dates and wanting to print t-shirts & host wiki's aside, There are 2 ways to approach making a live DVD: taking an install DVD & modifying it to the point that it runs dock & finder or some substitute (kismus, kalyway make bootable) , and taking a complete install and adding the ramdisk features etc to allow it to boot (modbin etc). I've made up 2 sticks which are essentially identical, except one has the minimal install image on it & the other has a retail DVD image. both are readwrite images so they can be modified and experimented with. A quick look seems to indicate that the retail DVD uses rc scripts whereas the installed OS uses launchd ? (take a look in /etc on the install dvd)
As the retail DVD image boots just fine, and I have used this type of stick a lot for toolkit purposes by launching stuff from terminal, clearly the ramdisk & booting setup from here can be used.
Also, there is much talk of "Live DVD" but little mention of what it actually is, so some goals should be agreed on regarding what it should be capable of etc. If a simple application launcher with a selection of utilities is all that is needed, then I'm pretty sure that could be done with a stripped-out install DVD, and should be possible to make cd-sized. or at least ~1gb
On the other end of the scale, if we say that the requirements are dock, finder, multiple applications running etc, that is a whole lot more work, and should be approached in stages.
Based on this, I think the first project should be to follow the modbin pattern and add the ramdisk features from the retail DVD to the installed image, and see how far that gets?
Just trying to be practical.
As the retail DVD image boots just fine, and I have used this type of stick a lot for toolkit purposes by launching stuff from terminal, clearly the ramdisk & booting setup from here can be used.
Also, there is much talk of "Live DVD" but little mention of what it actually is, so some goals should be agreed on regarding what it should be capable of etc. If a simple application launcher with a selection of utilities is all that is needed, then I'm pretty sure that could be done with a stripped-out install DVD, and should be possible to make cd-sized. or at least ~1gb
On the other end of the scale, if we say that the requirements are dock, finder, multiple applications running etc, that is a whole lot more work, and should be approached in stages.
Based on this, I think the first project should be to follow the modbin pattern and add the ramdisk features from the retail DVD to the installed image, and see how far that gets?
Just trying to be practical.
#55
Posted 31 October 2008 - 10:56 PM
Ok Hagar, just trying to contribute what I can.
I agree with Hagar on that we should create an outline, and then go from there.
I believe a limited os would be best. Easy to see how it runs on your system, the basic layout for new users. I don't think people need to run applications, that could introduce problems, etc.
I agree with Hagar on that we should create an outline, and then go from there.
I believe a limited os would be best. Easy to see how it runs on your system, the basic layout for new users. I don't think people need to run applications, that could introduce problems, etc.
#56
Posted 31 October 2008 - 11:50 PM
I agree with Hagar...
My idea is how about we start slow and try and get a limited bootable system and see how it goes. If it actually is snappy then we should keep going and turn it into a full environment over time.
How far do we want this project to go?
My idea is how about we start slow and try and get a limited bootable system and see how it goes. If it actually is snappy then we should keep going and turn it into a full environment over time.
How far do we want this project to go?
#57
Posted 01 November 2008 - 01:56 AM
Yeah. MoC is right. I think this project should start off limited, then if it is possible we could got to a full system. maybe a boot key where you could have it boot into limited mode, otherwise into full when we get that far.
what do you mean by how far?
what do you mean by how far?
#58
Posted 01 November 2008 - 02:22 AM
skyhighmac, on Oct 31 2008, 09:56 PM, said:
Yeah. MoC is right. I think this project should start off limited, then if it is possible we could got to a full system. maybe a boot key where you could have it boot into limited mode, otherwise into full when we get that far.
what do you mean by how far?
what do you mean by how far?
I meant this as do we want it to be more than a minimal system? Or do we want it to be something fully blown?
We should be thinking about this to know how to proceed. But the whole start out small go out big is a good idea.
#59
Posted 01 November 2008 - 02:24 AM
Well I think a combo, so (in the future) it would boot automatically into full featured mode, and then during boot, you can press a key and boot into minimal.
But yes, starting out small will be the best, and we already have the groundwork set up by Hagar and a few others...
But yes, starting out small will be the best, and we already have the groundwork set up by Hagar and a few others...
#60
Posted 01 November 2008 - 03:32 AM
I agree with the small-to-big approach, especially cause we'll never get something working otherwise
...but all that said, we should not try to pile hacks on to the small setup when we move bigger. We'll definitely need to switch approaches some when we move to a more full-featured setup. I really would like to avoid hacking up an install image to make it full-featured when we want finder and all the normal features...
...but all that said, we should not try to pile hacks on to the small setup when we move bigger. We'll definitely need to switch approaches some when we move to a more full-featured setup. I really would like to avoid hacking up an install image to make it full-featured when we want finder and all the normal features...
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users



Sign In
Create Account








