On Mon, 2024-08-12 at 18:06 -0400, Benjamin Marzinski wrote: > On Mon, Aug 12, 2024 at 06:44:54PM +0200, Martin Wilck wrote: > > On Fri, 2024-08-09 at 14:14 -0400, Benjamin Marzinski wrote: > > > > > > Despite some small misgivings about renaming symbols (which quite > > > possibly aren't being used by anyone) in our user-facing > > > libraries, > > > libpmathpersist and libmpathcmd, for the set: > > > > As the functions are only renamed, we could use (weak) aliases here > > to > > still provide the old function names if some caller requires them. > > Would you prefer that? > > > > We'd still be violating the glibc standards somewhat, but the > > probability for an actual symbol name clash is very low, so I guess > > it > > would be acceptable to do this in the name of backward > > compatibility. > > While it's not that big of a deal, I do think we should prioritize > backwards compatibility over this level of standards compliance. Ok. Sending patches right now. I've given them quick tests with qemu- pr-helper and some trivial test program that use the "old" ABI, but I'd be grateful if you could double-check. Martin
