On May 20, 2010, at 1:01 PM, Nicolas Williams wrote: > On Thu, May 20, 2010 at 12:18:58PM -0700, Don Cragun wrote: >> The reason that LOGNAME_MAX was stuck at 8 in <limits.h> for so long >> is that the System V ABIs and the SCDs require that value. >> >> Solaris 10 has been breaking ABI requirements around the edges for a >> few years. Since this is case is departing from more ABI >> requirements, should it have a major release binding? Or, should an >> opinion be written for this case acknowledging that the ARC knows >> that this case violates the ABIs and that the decision to do so is >> intentional (without setting precedent to otherwise ignore the ABI)? >> >> Once upon a time, there was a gang of four working on a >> definition of what would be the limits of the changes going into >> "Solaris next", whether it would be classified as a major or minor >> release, and what would constitute the basis for determining whether > > IMO the 8 byte limit on usernames is so onerous and obnoxious that > regardless of anyone might classify this change's release binding (Major > or Minor), the change has got to be made. Even 32 bytes is too short, > but the utmpx ABI breakage issues seem more severe (or at least complex) > than all the other ABI issues that might arise from going from 8 to 32 > bytes.
I'm not disagreeing with the move to 32 bytes. I just believe that the ARC needs to make it clear that doing so is a conscious decision to break the ABIs and that it does not set a precedent for other ABI breakage. If I remember correctly, an opinion needs to be written to do that even if this is just a fast track case. > > The _POSIX_LOGIN_NAME_MAX and LOGIN_NAME_MAX getconfs could be made > configurable, but given that there's not much that can be done about > determining LOGNAME_MAX and L_cuserid at run-time, the last two can and > must only be changed to allow for 32-byte usernames. > > In any case, customers that require strict SysV ABI compliance (e.g., > customers that have apps that use LOGNAME_MAX and/or L_cuserid and who > cannot or will not re-build those apps) can always stick to creating > usernames with 8 or fewer bytes. Yes, there could be a configuration > setting so that useradd knows to enforce 8 byte usernames in that case, > but such a thing seems rather useless to me, and the default had better > be 32 bytes anyways. > >> or not an implementation of OpenSolaris would be able to use the >> Solaris trademark. [...] > > (Did you mean the UNIX trademark?) No. The Open Group controls the definition of what implementations must do in order to use the UNIX trademark. To enforce its control over the trademark, The Open Group publishes the Single UNIX Specifications, creates test suites to be used as an indicator of conformance to the SUS, requires documentation for all behavior described by SUS as "implementation-defined", requires timely fixes for non-conforming behavior detected by TOG or by users, charges fees for administration of the branding program, etc. Once upon a time, Sun was planning to do the same thing with the Solaris trademark and/or the OpenSolaris trademark. This part of this discussion is probably off-topic for this case, but I would like the ARC to address this issue at some point in a future OpenSolaris ARC meeting as a topic on the agenda. - Don > > Perhaps > > Nico > -- > _______________________________________________ opensolaris-arc mailing list [email protected]
