| 1 |
On 2020-11-14 23:12, James Le Cuirot wrote: |
| 2 |
> I'm not claiming this has never actually happened but I use these |
| 3 |
> GitHub tarballs*a lot* and I don't recall ever seeing it. Does anyone |
| 4 |
> know for sure that it's happened in, say, the last 3 years? It's a lot |
| 5 |
> of extra work for a problem that may no longer exist or is so rare that |
| 6 |
> it's just not worth the effort. |
| 7 |
|
| 8 |
I am aware of three incidents: |
| 9 |
|
| 10 |
In 2011 but I cannot find details at the moment. |
| 11 |
|
| 12 |
In 2013, a bugfix in git |
| 13 |
(https://github.com/git/git/commit/22f0dcd9634a818a0c83f23ea1a48f2d620c0546) |
| 14 |
caused such a change. |
| 15 |
|
| 16 |
In second half of 2017, GitHub was rolling out a GZIP update across |
| 17 |
their fleet which caused such a change. It mostly hit rarely downloaded |
| 18 |
packages which were cleaned from CDN so tarball had to be re-generated. |
| 19 |
|
| 20 |
You can use GitHub Actions for example to automate this or include it in |
| 21 |
existing release workflows. But yes, you have to get upstream's |
| 22 |
attention to implement this. |
| 23 |
|
| 24 |
And it's not just GitHub, don't forget about GitLab and those |
| 25 |
self-hosted GitLab instances which often don't support to upload |
| 26 |
arbitrary assets... |
| 27 |
|
| 28 |
|
| 29 |
-- |
| 30 |
Regards, |
| 31 |
Thomas Deutschmann / Gentoo Linux Developer |
| 32 |
C4DD 695F A713 8F24 2AA1 5638 5849 7EE5 1D5D 74A5 |
Archives