| 1 |
Crystalsun <crystalsun@××××××××××.com> writes: |
| 2 |
|
| 3 |
> Anyone thought about implementing Emerge in Common Lisp? |
| 4 |
I did. I know some other people did, too. |
| 5 |
|
| 6 |
> I guess it could be useful due to the interactiveness Common Lisp |
| 7 |
> provides |
| 8 |
Agree. |
| 9 |
|
| 10 |
> if emerge was written in Common Lisp, it could be possible to |
| 11 |
> interactively debug the error and continue the process |
| 12 |
Not neccesarily. ebuilds are still ebuilds, namely Unix shell |
| 13 |
scripts. Most of my problems come from them not from emerge per se. |
| 14 |
|
| 15 |
> I know it may sound ambitious |
| 16 |
It does. |
| 17 |
|
| 18 |
> but at least it seems like an interesting idea to me, would be great |
| 19 |
> if anyone has thoughts on this topic |
| 20 |
Portage is updated regularly. EAPI evolves. Try estimating how many |
| 21 |
man-hours are put into it. That will give you some perspective on |
| 22 |
man-hours needed to maintain another implementation. Make some research |
| 23 |
on existing attempts (Paludis). |
| 24 |
|
| 25 |
The most promising way in my amateur-ish opinion would be to localise |
| 26 |
some task which Portage is slow at, and try to reimplement it (100% |
| 27 |
correctly!) for immediate and hopefully long-lasting gains. At least |
| 28 |
that could provide the project with some momentum and attention early |
| 29 |
on. |
| 30 |
|
| 31 |
Another attractive goal would be the ability to resolve |
| 32 |
dependencies interactively. I'm not even where to start with this |
| 33 |
though. Maybe if user could increase the backtracking threshold |
| 34 |
(emerge --backtrack=n) without restarting the whole thing, that would |
| 35 |
already be worth it, who knows. That would require lazy search of some |
| 36 |
kind. |
Archives