Originally posted by erniv2
View Post
Technically, chances are that one could:
* instrument a build system + toolchain + set of headers to call the BOINC API at select places for providing progress information, poking the watchdog, etc.
* package the aforementioned components as a BOINC application, hopefully in a modular way;
* sign and package source code bases as BOINC Work Units.
Of course, some important matters, such as trustworthiness of built binaries, and ensuring that there's a sufficient quorum of computers not controlled by the same group of potentially malicious actors running the WUs, would have to be tackled beforehand Otherwise it's complexity for little benefit.
However, IIUC, the bottleneck does not lie in build automation for 32-bit x86 packages, or computing power to perform them; it's more a matter of finding humans able and willing to spending time importing new releases into the build system, adjusting the packaging for the new releases, fixing the code or build steps when something breaks on a 32-bit architecture, testing the resulting packages, etc. - despite fewer and fewer computers and their human rulers benefiting from said work, as time passes by.
From the POV of an unpaid maintainer working on one's free time, there must be some things which have higher impact for real-world users than attempting to help maintain large sets of 32-bit packages, "large" here meaning "larger than somewhat beyond the set of packages needed by 32-bit x86 Linux / Windows games and game platforms".
Comment