Ian Collins wrote: > Roland Mainz wrote: > >[EMAIL PROTECTED] wrote: > > > >>>Is it possible to extend that to 12 months, please ? > >>>Some of the larger projects may have to wait longer for their inclusion > >>>into OS/Net and IMO it may be bad if the original B[1-9][1-9] build > >>>tools, sources etc. go away shortly before the putback just because > >>>they're slightly over the six-month barrier... > >>> > >>I don't understand that requirement. Any project needs to be build > >>against the latest bits prior to putback. So keeping exceedingly stale > >>bits does not help them in anyway. > > > >It depends on what you are working on. People may want to remain with > >their "older" version as long as possible because an "update" costs > >significant time, including such "small" items like installing a machine > >from scratch with a new version of OpenSolaris. Within a suiteable > >environment with multiple machines (+ Jumpstart etc.) this is easy and > >painless - but for contributors with less infratructure (and/or less > >experience) this can take significant time, sometimes half a day, a day > >or even much longer if problems with the hardware arise. > > > Isn't this time well spent? Working on stale code can cause all sorts > of integration problems later on.
The time is "well spend" when the people need it. FORCING people to update is bad. For example I've hit enougth problems since B37 with my MaxData laptop (native hardware) and VMware to make me complain about the time I've spend with jumpstarting machines over and over again. And I am working with Solaris (and before that with SunOS) for many many years. Now imagine how a 2nd-semester student (and Solaris-rookie) may react when he has to update his working machine every two months. > I'm a firm believer in continuous integration, especially when several > people are working on the same source. It depends on the viewpoint. On SPARC this is likely less a problem as the amount of hardware is limited. On x86 you have much more different hardware which may make such "updates" a serious pain. At least in my case we first tested which OpenSolaris build was working for everyone before we did the update - we had the choice between B35, B36, B37 and B38 and finally selected B37 because it was the only build which was working for everyone on all plaftform combinations needed. A "forced" update is very bad in such a case since multiple OpenSolaris builds need to be installed and evaluated and it's a dumb and thankless work. And it costs much time, a resource which is very limited (at least for people who have to work for food on other stuff, e.g. who aren't paid to hack Solaris the whole day...). ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) [EMAIL PROTECTED] \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 7950090 (;O/ \/ \O;) _______________________________________________ opensolaris-discuss mailing list [email protected]
