On Fri, Feb 20, 2009 at 10:11 AM, Jennifer Pioch <piochjennifer at googlemail.com> wrote: > On 2/20/09, Alan Burlison <Alan.Burlison at sun.com> wrote: >> Gavin Maltby wrote: >> >> >> > That won't eliminate issues such as the execname issue for pgrep, >> > but we can confirm it's being about as honest and clean as >> > a shell script can be. >> > >> >> If we are replacing utilities like sleep(1) then we should be using the GNU >> versions not the ksh93 versions, > > Now the great Alan Burlison comes and declares the work largest > Opensolaris team outside Sun as unnecessary. Sounds I've wasted the 2 > years here and should directly go to Linux. At least they honor work > by contributors. > Thank you.
Let's calm down a bit. I personally disagree with 'Linux compatibility' being the bar we should strive to (I think it's a can't win position to be in), but I'd rather try to convince others of my point of view and have a discussion to understand the thought processes and try not to get overly emotional (which is how I hope my emails have been taken, if not I apologize). >> not least because the GNU versions increase >> compatibility with Linux. > > What do people migrating from FreeBSD, HPUX or OSX in your > GNUGNUGNUfartlaris world? How to you handle POSIX incompatibilities? > Linux compatibility is not always a good thing if you alienate > enterprise users. I do think that's sometimes forgotten. Attracting more customers is always good, but want to be sure you're not doing it at the cost of those that are paying you money today. From what I've been told (unfortunately this is all 3rd party, so could be 100% false) that a last some the GNU maintainers are only interested in supporting the lowest common denominator (thus features that might differentiate on OS from another are not supported). Also, I'm not sure the compatibility concerns (which I think are a plus for Opensolaris over other OSes) are as high a priority to them as it is to us (which not a bad thing, but just is a reasonable difference in priorities). It's no always obvious (to me at least) that those things are always taken into consideration, especially when a lot of times the amount of effort required to achieve 99% of the compatibility is so trivial. For example, making Solaris ls 99% compatible with GNU ls and adding grep -r support is about 1000 lines of changes written and tested in a few hours scattered across a couple of days.
