1 |
On Wed, Aug 04, 2004 at 08:34:24AM +0900, Jason Stubbs wrote: |
2 |
> On Monday 02 August 2004 06:27, Martin Schlemmer wrote: |
3 |
> > On Sun, 2004-08-01 at 04:01, Jason Stubbs wrote: |
4 |
> > > On Sunday 01 August 2004 01:29, Mike Frysinger wrote: |
5 |
> > > > On Saturday 31 July 2004 04:32 am, Jason Stubbs wrote: |
6 |
> > > > > Well, the idea was that to create local USE flags of the form |
7 |
> > > > > cc-${ARCH}, have the gcc ebuilds use them to configure the cross |
8 |
> > > > > compiler and also have them control the the slot. The only real |
9 |
> > > > > difference to what there is now would be that portage would be able |
10 |
> > > > > to forecast the SLOT correctly. |
11 |
> > > > |
12 |
> > > > so the user would do USE="cc-arm" to get an arm cross compiler ? |
13 |
> > > > |
14 |
> > > > and then in the ebuild we'd have to do something like: |
15 |
> > > > SLOT="cc-arm? ( blah )" ? |
16 |
> > > > how'd would we specify a default slot ? |
17 |
> > > |
18 |
> > > SLOT="!cc-arm? ( !cc-foo? ( !cc-bar? ( !cc-baz ( defaultslot ) ) )" |
19 |
> > |
20 |
> > I really do not know why there hasn't been a '&&' and '||' logic |
21 |
> > added yet? Will it really be that difficult to implement? |
22 |
> |
23 |
> It would take some time to implement and then testing time and what not. The |
24 |
> main reason it hasn't been done is that it doesn't add anything that can't be |
25 |
> done already. |
26 |
> |
27 |
> > > Terrible, no? This is gone anyway. |
28 |
> > |
29 |
> > What exactly do you mean by this? |
30 |
> |
31 |
> There is no point adding complexity to SLOT (and making dep calculation |
32 |
> intricate and very slow) for the benefit of four ebuilds that could be |
33 |
> implemented without the use of dynamic SLOTs anyway. |
34 |
> |
35 |
> As far as I know, allowing ${KV} as a SLOT was a quick hack to prevent |
36 |
> previously installed kernel modules from being unmerged. It was/is a |
37 |
> temporary solution until something better is found. That better solution is |
38 |
> what I'm looking for here. |
39 |
> |
40 |
> Regards, |
41 |
> Jason Stubbs |
42 |
> |
43 |
|
44 |
I'm happy those issues are being discussed as i didn't have much incentive to |
45 |
post a feature request like what follows until now. First, some context: |
46 |
I maintain the ML packages and particularly the ocaml ones. As some of |
47 |
you may know, ocaml is not binary compatible from version to version and |
48 |
it is required to recompile all dependent libraries on compiler updates, |
49 |
(in the right order of course). This situation is very similar with Perl |
50 |
or Python actually. Managing more than one version of such packages is |
51 |
currently impossible (using portage only) AFAIK. |
52 |
|
53 |
What I would like is a proper framework to deal with this |
54 |
compilers/interpreters, and hopefully one general enough to deal with |
55 |
cross-compilation too. I have hacked a solution on my own to deal with |
56 |
the problem of incompatible compilers, using dynamic SLOTS, an ocaml |
57 |
eclass and an ocaml-config script (similar to gcc-config). All ocaml |
58 |
packages use this eclass which sets the slot to the compiler version and |
59 |
sets some env vars (pathes) to allow flawless cohabitation of the |
60 |
different compiler versions. It works just like a charm. I can update |
61 |
all my packages for new versions of ocaml, do backward-compatibility |
62 |
tests (i'm an ocaml dev) and if i wanted to, i could manage the CVS |
63 |
version of ocaml and all ocaml packages using only portage and my |
64 |
ocaml-config script. |
65 |
|
66 |
The problem is obviously that i'm using portage for something it was not |
67 |
intended to do (cleanly, at least). I have some (clear?) ideas about how it |
68 |
could be done though, and i'm willing to give help. Here's my wish: |
69 |
|
70 |
*** Portage support for compilers *** |
71 |
|
72 |
I never heard anybody mention it, but I think it is required to get a |
73 |
clean implementation. It would introduce some changes: |
74 |
|
75 |
- Factorized, standard config files and utilities for all compilers/interpreters, |
76 |
gcc, python, perl, ocaml, you name it. Configurations would be simply a set |
77 |
of env vars, much like what gcc-config uses today. |
78 |
You could set different configurations for different users etc... |
79 |
A set of compiler policies (in the tree) would also be needed to |
80 |
mention the fact that no ocaml compilers are compatible, but that |
81 |
different gcc's are. A new flag could be added to emerge telling him |
82 |
to rebuild automatically all packages that depended on a previous |
83 |
compiler that is incompatible with a newer one. |
84 |
|
85 |
- An ebuild metadata variable mentioning the compiler configurations |
86 |
it depends on: a CDEPEND in fact, mentioning which compiler versions |
87 |
it is compatible with. Unset would mean any, but when set, it could |
88 |
well be cached and maybe some pruning in the dep tree could be done too. |
89 |
|
90 |
- When invoking an ebuild for some package, portage would simply set |
91 |
the compiler variables (for the compilers mentionned in CDEPEND). |
92 |
The process of dependency checking would introduce one more step on |
93 |
each depended-on pkg check : When a package dependent on the same language |
94 |
as one mentionned is found, try to find an installed one matching the |
95 |
dep requirements and the compiler policies, if none matches, use the |
96 |
same policies as the current one for dep-checks etc... Informative |
97 |
messages could be easily crafted in case of errors. |
98 |
|
99 |
Now for the critics, i will point out that my scheme doesn't extend |
100 |
ebuild syntax, which is good as most compiler compatibility problems can |
101 |
be handled at once and for all packages. It would also factorize some |
102 |
code that's between the various *-config scripts and as an added benefit |
103 |
it would make every compiler update easy, getting rid of rebuild |
104 |
scripts. The main advantage is a finer, more robust dependency |
105 |
checking. |
106 |
|
107 |
Here's a test case: |
108 |
I've got ocaml-3.07, lablgl and lablgtk merged, i want the latest |
109 |
mldonkey which requires lablgtk-1.2.7 which in turn works only with |
110 |
ocaml-3.08. Portage will see that mldonkey has no particular preference |
111 |
regarding compiler versions, it will check lablgtk which will tell that |
112 |
it needs ocaml-3.08 and no other, so, as you have the opengl USE flag |
113 |
set, portage will check ocaml compiler policies to see if lablgl in 3.07 |
114 |
would be compatible with lablgtk compiled with 3.08, the policy says no, |
115 |
hence portage checks that lablgl can be compiled with 3.08 and finally |
116 |
it sets its roadmap to: install 3.08, lablgl for 3.08, lablgtk for 3.08 and |
117 |
finally merge mldonkey for 3.08. |
118 |
|
119 |
Now for the cross-compiling, it would greatly help to have such support |
120 |
built into portage. If i want to build ppc packages on my athlon, |
121 |
i can tell portage to use a (more general than just compilers) |
122 |
configuration that's declared incompatible with my current one |
123 |
(as it should be) that sets the root to some other place, have a |
124 |
particular make.conf, cross-compiler config etc... I'm just prospecting here. |
125 |
|
126 |
That was just my 2 euros, i'm more than open to criticism about my |
127 |
proposal which i ruminate for some time now (see bug #39246), and |
128 |
i'll get it as far as I can, writing a GLEP and coding included (i'm in |
129 |
holidays after all :). |
130 |
|
131 |
Thanks for your patience ! |
132 |
|
133 |
-- mattam |