| 1 |
On Tue, 2021-07-27 at 21:18 +0100, Neil Bothwick wrote: |
| 2 |
> |
| 3 |
> Instead of continually beating on portage on this list, which will |
| 4 |
> achieve nothing more than a minor waste of electrons, you should be |
| 5 |
> focussing on getting the ebuilds fixed so that portage is no longer given |
| 6 |
> conflicting or incorrect information. |
| 7 |
|
| 8 |
In most cases this is good advice. The problem with djbdns/qmail is |
| 9 |
that they are abandoned upstream, and kept alive in Gentoo only by a |
| 10 |
patchwork of... well, patches. |
| 11 |
|
| 12 |
It would not -- independently -- be too much work to fix either package |
| 13 |
to work without daemontools. But, since they are abandoned, there is |
| 14 |
nowhere to send the fixes. That would add yet another patch to both |
| 15 |
packages. Qmail is similar, but I know more about djbdns, so: for |
| 16 |
example, net-dns/djbdns already applies SEVENTEEN PATCHES. And many of |
| 17 |
them are quite large. |
| 18 |
|
| 19 |
When a new security issue is found and a new patch is created or an |
| 20 |
existing one changes, then all of a sudden we get conflicts. If we |
| 21 |
apply the new one first, then (say) patches two through five might |
| 22 |
apply cleanly, but patches six through eighteen will fail. Now we have |
| 23 |
to manually rebase thirteen patches? That just will not happen, which |
| 24 |
is why no one has fixed these two packages to work without daemontools |
| 25 |
yet. The cost of additional patching is too high. |
| 26 |
|
| 27 |
You should really just avoid both of them. This is an obscure problem |
| 28 |
because no one chooses either djbdns or qmail since 2005. |
Archives