| 1 |
On 02/04/2018 06:14 PM, Canek Peláez Valdés wrote: |
| 2 |
> |
| 3 |
> |
| 4 |
> What would this even mean in the context of a source-based distro? |
| 5 |
> |
| 6 |
> |
| 7 |
> It would mean that we all could reproduce the exact same bugs given |
| 8 |
> the CFLAGS/USE/etc. combination. |
| 9 |
> |
| 10 |
> Many groups are working on this from different fronts; if the results |
| 11 |
> stabilize at some point, Gentoo could use that to at least give the |
| 12 |
> users the option of enabling reproducible builds. |
| 13 |
+1 |
| 14 |
|
| 15 |
That was exactly what I found interesting. |
| 16 |
There is also some best practices that are pointed in the documentation |
| 17 |
that seems to be useful for achieving the deterministic build: |
| 18 |
https://reproducible-builds.org/docs/ |
| 19 |
|
| 20 |
Gentoo is a source based distro but the build environments and profiles |
| 21 |
are many times the challenge that trigger most of bug reports in bugzilla. |
| 22 |
With the reference reproducible build for some ebuild or profile would |
| 23 |
allow to review the details of portage configuration that could be the |
| 24 |
cause of the compile failures. This would be very useful for keyword |
| 25 |
masked packages and for those that desire to be bleeding edge. Also for |
| 26 |
overlays that would be qualiity assured by the reproducible builds over |
| 27 |
EAPI verification. |
| 28 |
|
| 29 |
I'm convinced that some of the ideas would be useful even for a souce |
| 30 |
based distribution. In a source distribution I think that this would be |
| 31 |
simple when comparing to binary distributions. All necessary information |
| 32 |
would be collected directly from emerge and published as is. |
| 33 |
|
| 34 |
Thanks to all for the feedback |
Archives