4 ms·
From 2019. Has there been any progress on the issues since then?
by kardos 6y ago
From 2019. Has there been any progress on the issues since then?
- wahern 6y agoThe answer, at least for Linux, was always to just use the pre-existing copy_file_range syscall. Patching GNU coreutils' cp(1) utility took less effort than either the LWN summary or the presentation it summarized: https://github.com/coreutils/coreutils/commit/4b04a0c3 https://github.com/coreutils/coreutils/commit/4b04a0c3 Alas, we live in an age where more engineers blog about open source software than actually contribute code :( But even worse is that major corporate users like Amazon simply decided to write corporate coding policies instead of addressing the underlying issue with some simple patches. Though, to be fair, it's not always easy--I was unable to cut through the noise on the netfilter mailing-list to get attention for a patch to fix a bug in the iptables utility. And a bug report with extensive analysis (that took days) which resulted in a fix to Ubuntu's Python + FIPS-compliant OpenSSL was regressed just a few months later :( I've been too busy to file another bug report because my company doesn't actually allot time for such fixes, either. In both cases I put in evening hours to try to fix the issue upstream, but those opportunities have been exceedingly scarce of late.
- kardos 6y agoThanks for the reply & pointing to the change. Hopefully there will be another stable coreutils release coming soon. Agreed that it behooves projects to make the contributing process straightforward...