| 1 |
On 23/03/06, Paul de Vrieze <pauldv@g.o> wrote: |
| 2 |
> |
| 3 |
> Certainly, |
| 4 |
> |
| 5 |
> chroot combined with lvm snapshots would be the easiest way. |
| 6 |
> |
| 7 |
> If you want to focus on binary packages, you might want to start with not |
| 8 |
> doing it automatically, but using some crude heuristics. You can make it |
| 9 |
> configurable for when the heuristics don't work. |
| 10 |
> |
| 11 |
> Basically what is the difference between a binary package and the source |
| 12 |
> package is the dependencies. |
| 13 |
> |
| 14 |
> You can regard a dependency as a restriction on the configurations in |
| 15 |
> which |
| 16 |
> the package can be installed. For binary packages only runtime |
| 17 |
> dependencies |
| 18 |
> are relevant (well in gentoo anyway, we don't have initial configuration |
| 19 |
> deps). Binary packages do however further restrict the configurations from |
| 20 |
> the runtime dependencies. These restrictions originate from the |
| 21 |
> configuration |
| 22 |
> when compiling the package. |
| 23 |
> |
| 24 |
> Then I have some good and bad news. The good news: |
| 25 |
> - Binary dependencies are as necessary for binary packages, as for |
| 26 |
> validating |
| 27 |
> a current configuration (set of installed packages). |
| 28 |
> The bad: |
| 29 |
> - For proper binary dependencies the depend expression possibilities must |
| 30 |
> be |
| 31 |
> extended. Useflag dependencies would be at least usefull. BINSLOT would |
| 32 |
> be |
| 33 |
> almost required. We already saw that SLOT can't always be used for |
| 34 |
> BINSLOT. |
| 35 |
|
| 36 |
|
| 37 |
BINSLOT is a new word for me. |
| 38 |
|
| 39 |
I dont see any meaning in calculating dependencies after build. There would |
| 40 |
be too much problems: |
| 41 |
* If any of packages you depend on, are masked, you get an error after |
| 42 |
build, but should get it before |
| 43 |
* In many cases, not only libs, but also headers are included from another |
| 44 |
pack [kernel headers, for example, as i have seen from error messages] |
| 45 |
* You cant pre-calculate dependancies |
| 46 |
|
| 47 |
My interest, for sure, is code dependancies. At first i see a clear way to |
| 48 |
get it work for now, secondly i think it can be used for bin deps, also. |
| 49 |
|
| 50 |
In cases, where both headers and libs are included from other pack, which is |
| 51 |
same on both cases, this is totally irrelevant for deps if this other pack |
| 52 |
is statically or dynamically used. |
| 53 |
|
| 54 |
In cases, where headers are in pack to be compiled and .so is in another |
| 55 |
package, i havent yet checked, how to understand if dependancy is binary -- |
| 56 |
i cant say for sure that this does not include some work from user. |
| 57 |
|
| 58 |
Ok, i better write few lines of code now as i know where i should start (or |
| 59 |
at least i have one part of code, which could be discussed here, in front of |
| 60 |
my minds-eye). |
| 61 |
|
| 62 |
Thanks. You will understand me better when something works :) |
| 63 |
|
| 64 |
Paul |
| 65 |
> |
| 66 |
> -- |
| 67 |
> Paul de Vrieze |
| 68 |
> Gentoo Developer |
| 69 |
> Mail: pauldv@g.o |
| 70 |
> Homepage: http://www.devrieze.net |
| 71 |
> |
| 72 |
> |
| 73 |
> |
| 74 |
|
| 75 |
|
| 76 |
-- |
| 77 |
tvali |
| 78 |
|
| 79 |
Measuring programming progress by lines of code is like measuring aircraft |
| 80 |
building progress by weight. -Bill Gates |
| 81 |
|
| 82 |
For every complex problem there is an answer that is clear, simple, and |
| 83 |
wrong. -H L Mencken |
| 84 |
|
| 85 |
Theory is when you know something, but it doesn't work. Practice is when |
| 86 |
something works, but you don't know why. Programmers combine theory and |
| 87 |
practice: Nothing works and they don't know why. |
| 88 |
|
| 89 |
http://www.eskimo.com/~hottub/software/programming_quotes.html |
| 90 |
http://www.softwarequotes.com/ |
Archives