| 1 |
On 29-09-2010 16:20:00 +0200, Peter Waller wrote: |
| 2 |
> This seems like bad practice not to have a way to do this. I don't think you |
| 3 |
> can reasonably expect everyone updating their prefix installation to see |
| 4 |
> posts on this mailing list. You're talking about causing people considerable |
| 5 |
> pain, possibly effectively irreversible - at least to less experienced |
| 6 |
> users - unless you publish the details of your hacks required to fix it |
| 7 |
> after it is broken. |
| 8 |
|
| 9 |
I can understand your concerns. Gentoo Prefix is still an experiment, |
| 10 |
though. While I don't really like to break things, I can't really avoid |
| 11 |
it always either. |
| 12 |
|
| 13 |
> I don't see why such breakage should be necessary in the circumstance that |
| 14 |
> someone innocently wants to update some packages. Isn't Jeremy's solution |
| 15 |
> possible, for instance? |
| 16 |
|
| 17 |
I had to mask any flawed portage version, which means most people will |
| 18 |
have to downgrade portage now, which I'm not going to force through |
| 19 |
openssl or any other random package. It's a very structural problem. |
| 20 |
|
| 21 |
The real problem is an abi update of a package, and Portage's |
| 22 |
preserve-libs not working. In our case even worse, because that means |
| 23 |
manual means to prevent huge breakage are disabled because Portage is |
| 24 |
supposed to handle it. |
| 25 |
|
| 26 |
|
| 27 |
-- |
| 28 |
Fabian Groffen |
| 29 |
Gentoo on a different level |
Archives