What we want is to be able to remotely access kernels and isolate them. But this quite difficult and a solution probably requires having extra bits.
Stef On Sep 11, 2011, at 2:49 AM, Schwab,Wilhelm K wrote: > Stef, > > It's lonely at the top :) Best wishes with your new opportunities. > > As a linux user, one thing I cringe about is opening a second image and > creating a COMPLETE mess of my image. The IDE really needs to detect the > problem and refuse to open before badness happens. Otherwise, I make backups > (not as often as I should) and good use of recovering lost changes. > > As for the general problem of building images, I see it much as lack of > tools. We need something (Migrate being my evolving solution) that looks at > what is installed and can get it into the next image. Metacello is a welcome > way to install complicated stuff from others, by *my* code is presumably > always something that I want to save and load w/o hassles. > > FWIW, I have a love-hate relationship with R, and that community is facing a > similar problem, and their presumed solution appears to be something that > will analyze and build. I realize R's language is awful (raises doubt), but > the system as a whole is quite impressive (restores faith). > > Bill > > > ________________________________________ > From: [email protected] > [[email protected]] On Behalf Of Stéphane Ducasse > [[email protected]] > Sent: Saturday, September 10, 2011 4:37 PM > To: [email protected] > Subject: Re: [Pharo-project] 1.3 OneCick with Cog VM > > On Sep 10, 2011, at 9:00 PM, John Toohey wrote: > >> I saw the screencast where Mariano did something like that. I had just >> built the image, when the original VM crash happened, so it wasn't >> that much of a deal to rebuild it. >> >> I would like to see some form of safe-mode boot for Pharo, so that >> just the kernel would be loaded. ST images are all about state, and >> when production servers go down, it would be nice to be able to start >> them up, and rewind the app. For normal apps this is not an issue, but >> when forking processes for socket IO or a FFI call, it is possible to >> bring it down (or in my case where I think my error handling code >> caused an error) , and moving in a backup image is not always enough. > > I would like to be able to rebuild the kernel at startup. > Now I should work on it but I got new responsibilities and they take my time. > > Stef >> >> I would like to get involved in such a project, if there was interest for it. >> >> On Fri, Sep 9, 2011 at 18:41, Sean P. DeNigris <[email protected]> wrote: >>> >>> John Toohey wrote: >>>> >>>> So do I have to rebuild a new image from scratch now? Is there any >>>> command line switch that I can use to by pass the loading of my app on >>>> start up? >>>> >>> >>> I don't know if it'd be worth it, but if you're talking about a lot of work >>> lost, it sounds like you could build a debug VM and bypass the loading code >>> from the debugger (like I just had to do). >>> >>> Sean >>> >>> -- >>> View this message in context: >>> http://forum.world.st/1-3-OneCick-with-Cog-VM-tp3798884p3802970.html >>> Sent from the Pharo Smalltalk mailing list archive at Nabble.com. >>> >>> >> >> >> >> -- >> ~JT >> > > >
