Re: [PATCH v3 2/3] ecc: prefix optimized ECC function names with underscore
пт, 10 янв. 2020 г. в 23:01, Niels Möller : > > Dmitry Eremin-Solenikov writes: > > > So did I at the time of writing a patch, finding no actual users of > > these functions. I think it is fine to drop them without bumping > > soname. > > It seems none disagrees with that. I've merged all three patches to the > master-updates branch for testing. Thank you! -- With best wishes Dmitry ___ nettle-bugs mailing list nettle-bugs@lists.lysator.liu.se http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
Re: [PATCH v3 2/3] ecc: prefix optimized ECC function names with underscore
Dmitry Eremin-Solenikov writes: > So did I at the time of writing a patch, finding no actual users of > these functions. I think it is fine to drop them without bumping > soname. It seems none disagrees with that. I've merged all three patches to the master-updates branch for testing. Thanks, /Niels Möller -- Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677. Internet email is subject to wholesale government surveillance. ___ nettle-bugs mailing list nettle-bugs@lists.lysator.liu.se http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
Re: [PATCH v3 2/3] ecc: prefix optimized ECC function names with underscore
Hello, вт, 7 янв. 2020 г. в 23:20, Niels Möller : > > dbarysh...@gmail.com writes: > > From: Dmitry Eremin-Solenikov > > > > There is no need to keep optimized ECC functions in public namespace > > (nettle_*), move them to internal namespace (_nettle_*). > > I agree this rename makes sense. > > But it could be considered an ABI break, since we remove symbols with > version string "HOGWEED_5", matching the soname, and those symbols are > supposed to be stable as long as the soname is the same. > > On the other hand, the symbols are not mentioned in docs or headers, and > they don't even exist in all configurations. So maybe it's ok to remove > them without changing the soname? That will break any application > needing them in libhogweed.so, but that application would break in the > same way if nettle were reconfigured with --disable-assembler. > > I had a try with > http://codesearch.debian.net/search?q=nettle_ecc_.*modp=0, and > at least the functions appear unused outside of Nettle. So did I at the time of writing a patch, finding no actual users of these functions. I think it is fine to drop them without bumping soname. -- With best wishes Dmitry ___ nettle-bugs mailing list nettle-bugs@lists.lysator.liu.se http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
Re: [PATCH v3 2/3] ecc: prefix optimized ECC function names with underscore
dbarysh...@gmail.com writes: > From: Dmitry Eremin-Solenikov > > There is no need to keep optimized ECC functions in public namespace > (nettle_*), move them to internal namespace (_nettle_*). I agree this rename makes sense. But it could be considered an ABI break, since we remove symbols with version string "HOGWEED_5", matching the soname, and those symbols are supposed to be stable as long as the soname is the same. On the other hand, the symbols are not mentioned in docs or headers, and they don't even exist in all configurations. So maybe it's ok to remove them without changing the soname? That will break any application needing them in libhogweed.so, but that application would break in the same way if nettle were reconfigured with --disable-assembler. I had a try with http://codesearch.debian.net/search?q=nettle_ecc_.*modp=0, and at least the functions appear unused outside of Nettle. Opinions? Regards, /Niels -- Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677. Internet email is subject to wholesale government surveillance. ___ nettle-bugs mailing list nettle-bugs@lists.lysator.liu.se http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
