Sivakumar Shanmugasundaram wrote: > > >> /usr/ruby/1.8/lib/ruby/1.8/i386-solaris2.11/dtrace.h > > > >Who is the consumer for this one? > > No one AFAIK. This is part of the patch and is consumed by the ruby > binary itself. The Joyent's patch had this file and when you run 'dmake > install', this file ends up in a sub-directory under the /usr/ruby/1.8 > directory. So does other .h files from Ruby source, irrespective of > whether it is consumed by others or not. The rubyinst.rb has logic, > where it installs ALL (*.h) header files to the target directory. > I was just being consistent with the existing Ruby ARC case, where all > such .h files were being listed. > In this case, dont you think this should stay?
So it's only relevant during build-time, there isn't really a need to ship it in the package? Or is there? If not, ideally it shouldn't be shipped. If the existing logic is such that all private .h files get copied anyway I won't object too much if this follows that convention. In any case don't document it in the ARC case since it looks like it is purely an implementation artifact. > We are yet to receive an official answer from Joyent on the stability. > Though it is not 'mature', I am reasonably confident that it wont be > changing very often to be considered. 'Volatile'. If you're confident, that's great. Uncommitted is certainly preferable, so keep it that way then. > BTW, any one knows how to add tables to man page (ruby.1) document? Look in the ATTRIBUTES section of most of the sfw manpages for a table sample. > >Please include the probe descriptions & arguments here (as well as in > >the man page) since it is vital info to being able to do anything with > >these probes. > > But wouldnt adding it here be a duplicate of the 'documentation' section > from above? Well, yes. But they're important interfaces so should be in the ARC case and they definitely need to be in the man page so users can conveniently find the info. > >Please add a section earlier in the functional spec document > >explaining the reasons why there is a difference between > >platforms. That's key info that needs to be documented, not just > >mentioned in a footnote. > > As you are aware, because of CR 6659110, the Sparc version of Ruby does > not have the probes. I was hoping that this will be resolved soon, > before the Indiana release, so I could fix the Sparc Ruby version too. > That way, I need to just update the man page only and not change the ARC > case. > Please let me know if this is okay or not. I do remember the history behind it but not every reader will know... Unless you are certain it's about to be fixed and you're willing to wait for that fix, better to just document the issue in the ARC case and explain the consequences. Then add that you'll fix the feature parity as a bug as soon as possible, whether next week or next year. -- Jyri J. Virkki - jyri.virkki at sun.com - Sun Microsystems
