In message <40F464DA.4030501_at_linuxpowered.com>, Jon Disnard writes: >>The correct solution is the add whatever it takes to make the world >>target fail if it is unsafe. >> >So possibly a system to parse UPDATING, or whatever, for ABI changes and >then halting the installworld stages of the world target? Well, maybe we need a different criteria for bumping __FreeBSD_version or maybe we need a parallel ABI version number, but parsing UPDATING would be a hack. >That seems good and bad, because sometimes people checkin without any >documentation, and this idea would absolutely require anything with a >potential to cause hazard to be documented with (or before) the >associated checkin. Are all the commiters diligent enough to do that? Is >there some other way we can have this idea of a "smart make world" that >acknowledges the case of lazy documentation? I think that if we make it a rule that __FreeBSD_version be bumped if make world without a kernel update would be unsafe would be followed. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk_at_FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.Received on Tue Jul 13 2004 - 20:46:28 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:01 UTC