| 1 |
Hey Ruud, |
| 2 |
|
| 3 |
Ruud Koolen <redlizard@g.o> writes: |
| 4 |
|
| 5 |
> In the mainline process, stage1 manually builds a handful or required tools |
| 6 |
> (such as make, patch, and python) in $EPREFIX/tmp, using the host toolchain. |
| 7 |
> stage2 then manually builds a portage and constructs a profile, both in |
| 8 |
> $EPREFIX. stage3 starts by using portage to build a toolchain in $EPREFIX |
| 9 |
> (along with minimal dependencies, such as bash and bison), built using the |
| 10 |
> host toolchain; afterwards, portage uses this toolchain to emerge a proper |
| 11 |
> portage (also including dependencies, like python), emerged on top of the |
| 12 |
> bootstrapped portage using USE=-collision-protect. At this point, portage is |
| 13 |
> basically operational, and what remains is emerge -e @system. |
| 14 |
> |
| 15 |
> [...] |
| 16 |
> |
| 17 |
> In my experimental process, things work differently. The stage1 part is |
| 18 |
> unchanged, but stage2 bootstraps a portage *in $EPREFIX/tmp*, and portage's |
| 19 |
> cross-eprefix functionality is used to have this portage build packages in |
| 20 |
> both $EPREFIX/tmp and $EPREFIX. In stage3, portage builds a toolchain (and |
| 21 |
> dependencies) in $EPREFIX/tmp, but one that targets $EPREFIX; that is, it |
| 22 |
> builds a toolchain in $EPREFIX/tmp that builds executables that expect to run |
| 23 |
> in $EPREFIX, have $EPREFIX for rpath, and read headers from $EPREFIX. This |
| 24 |
> toolchain is then used to build a second toolchain in $EPREFIX (targeting |
| 25 |
> $EPREFIX as normal); this "real" toolchain is then used to build the final |
| 26 |
> portage, in $EPREFIX. At this point, the situation is similar to the end of |
| 27 |
> stage3 in the mainline process, with an operational and self-sufficient |
| 28 |
> toolchain and portage in $EPREFIX, and all that is left to be done is emerge - |
| 29 |
> e @system. |
| 30 |
|
| 31 |
I'll omit your explanations which I am aware of: It's a nice wrap up |
| 32 |
indeed. So I'm going to cut in here. |
| 33 |
|
| 34 |
The new stageX feels intertwined: |
| 35 |
|
| 36 |
stage2: tmp portage |
| 37 |
stage3: tmp gcc, EPREFIX gcc, EPREFIX portage |
| 38 |
|
| 39 |
How about defining stage2 as "anything to be built with tmp portage into |
| 40 |
tmp", and stage3 as "anything to be inserted into EPREFIX"? Then it's |
| 41 |
arraged as: |
| 42 |
|
| 43 |
stage2: tmp portage, tmp gcc |
| 44 |
stage3: EPREFIX gcc, EPREFIX portage |
| 45 |
|
| 46 |
|
| 47 |
On the resistence from the momentum of tradition, I hope you can look at |
| 48 |
the situation from Fabian's angle of view: the fragile script (bugs |
| 49 |
sometimes) at least works on AIX, HPUX, solaris, MacOSX. They might be |
| 50 |
broken by the new bootstrap concept. As you don't have access to these |
| 51 |
platforms, several people have to be involved to prove the platforms are |
| 52 |
cool with the new concept, which implies a lot of work and social |
| 53 |
coordination than you might anticipate. Unless the present solution is a |
| 54 |
real pain in the ass, we are conservative with the refactorization. The |
| 55 |
new one is good enough and it's a good job, while it's counterpart |
| 56 |
should be bad enough to drive your fellows on track. |
| 57 |
|
| 58 |
If, by all means, you got development boxes of AIX, solaris, etc. And |
| 59 |
prove the new one can 100% replace the old one. That's when it starts |
| 60 |
getting widely accepted. At the same time, however, this process will |
| 61 |
take a lot of work. If it is your destiny, go for it. |
| 62 |
|
| 63 |
Cheers, |
| 64 |
Benda |
Archives