> On 2011-05-02 12:51:56, Nathan Binkert wrote: > > SConstruct, line 247 > > <http://reviews.m5sim.org/r/668/diff/1/?file=12211#file12211line247> > > > > Should we prompt the user for permission? That way the user would at > > least know that something happened. > > Steve Reinhardt wrote: > I doubt most people care. How many would really say no? Most people > just want things to be automatic. > > If they really do care they can always edit it out later (the comments > will let them know where the hooks came from). > > > Nathan Binkert wrote: > I agree that nobody would say no, but most people also would never notice > that it had happened because the message would just fly by if there was no > prompt. I guess that you're arguing that that would be a good thing :) > > Gabe Black wrote: > Changing config files behind peoples back is a really bad idea in my > opinion. I know I've stopped using entire distros (Suse) because they mucked > with config files behind my back and perpetually broke my system. My configs > are mine, and the minimal level of respect for that would be if we asked > permission before we let ourselves in. I think leaving well enough alone and > having the user go in and fix it themselves is actually the best approach. > > Nathan Binkert wrote: > This isn't a config file in your homedir, it's the config file in the M5 > repo and it will only be changed if you don't have it set up right, so it's > not all that bad. I have the same leaning that you do, so I think I'd rather > see a prompt for the user, but I don't feel that strongly about it. > > Steve Reinhardt wrote: > Yea, I would feel differently about ~/.hgrc, but this is just a local > config file, and it's one that will need updating each time you clone a new > repo. I don't mind making the warning a little more prominent, but since the > status quo is that we won't even try to compile if the hooks aren't there, it > seems overkill to me to make too big a deal of it... is someone going to be > so offended about this that they're going to go try a different simulator? > Plus if we could find a way to get these hooks to be part of the m5 repo so > that they were automatically in place whenever you cloned, we would have done > that, so I don't see how this is much different.
I'm just afraid of this happening, causing some problem, somebody spending a long time and discovering (or not discovering) that some config file was changed without their knowledge and that's what was breaking everything. I can't imagine somebody not being a little miffed about that. But if you think it's really really safe and you've made it obvious what happened then I guess it could be ok. Maybe make it stop scons after that warning prints if those lines aren't there so it's harder to miss? Then they can just run it again and it'll go through, but they at least know what happened. And yes, I have heard of people that have tried to use M5, got frustrated and gone elsewhere. It is very possible to be too annoying or difficult and to drive away users. - Gabe ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.m5sim.org/r/668/#review1173 ----------------------------------------------------------- On 2011-05-02 12:34:56, Steve Reinhardt wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.m5sim.org/r/668/ > ----------------------------------------------------------- > > (Updated 2011-05-02 12:34:56) > > > Review request for Default, Ali Saidi, Gabe Black, Steve Reinhardt, and > Nathan Binkert. > > > Summary > ------- > > SConstruct: automatically update .hg/hgrc with style hooks > > Seems easier than pestering people about it. > Note also that path is now absolute, so you don't get errors > when invoking hg from subdirectories. > > > Diffs > ----- > > SConstruct 66a3187a6714 > > Diff: http://reviews.m5sim.org/r/668/diff > > > Testing > ------- > > > Thanks, > > Steve > > _______________________________________________ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev
