Well, if we're dreaming anyway: My dream is to interact with a remote git repo. Because at the moment, to use Pharogenesis, one has to clone the repository and keep in mind to include all the tags and branches. It would be cool to have the network protocols integrated into GitFS. But that's a whole project on its own.
Thanks for the interest anyway! Feels good to do something that might actually be used or lead to a real change. Max On 17.01.2011, at 17:35, Norbert Hartl wrote: > > On 17.01.2011, at 16:41, Max Leske wrote: > >> I'm not quite sure I can follow you. Do you want to import the full >> repository into the image? Or would you want SystemDictionary to hold simply >> the version of every method that was active *before* the current version? >> >> The first case is a no go I think. The packed repository is 101 MB big and >> there are still 11000 object files in addition to the pack and index file. >> The second case would definitely speed up the look up process but on the >> other hand it doesn't solve the problem that you need another mechanism in >> the background (like a DB) to hold further versions. >> >> Was that even close to what you wanted? >> > I thought about the first one. I didn't think about resource restrictions at > first. Just that the best case IMHO would be to have everything at hand > extending only the MOP/query calls by a version dimension. It would then be > possible to instantiate old objects and play with them. > But then I was just day dreaming and it has less to do with what you are > doing :) > > Norbert >> >> On 17.01.2011, at 10:25, Norbert Hartl wrote: >> >>> >>> On 16.01.2011, at 21:46, Stéphane Ducasse wrote: >>> >>>> I would love to have all these versions in a database and that we can >>>> query them :) >>>> >>> I thought the same. But then I didn't think about database. Wouldn't that >>> be possible if the environment SystemDictionary is in full use? Would it be >>> possible to import a version into a SystemDictionary that reflects one >>> version and that is accesible through the class via class history?. Or is >>> this a rather stupid idea? >>> >>> Norbert >>> >>>> On Jan 16, 2011, at 12:26 PM, Max Leske wrote: >>>> >>>>> Hi Geert >>>>> >>>>> Pharogenesis is my bachelors thesis and I'm still not finished writing >>>>> it. I'll let you know when I'm finished. >>>>> >>>>> In short: I'm hosting a git repository on github that contains the >>>>> information from the changes and sources files starting at Squeak version >>>>> 1 up to Squeak 4.0 and Pharo 1.0. The idea is to have a single point of >>>>> entry to that information so that one could for instance find out about >>>>> every change to a method back to its creation. >>>>> >>>>> To accomplish this I built an implementation of git called GitFS that >>>>> uses Filesystem (it's based on GitForSqueak by Tony Garnock-Jones). You >>>>> can find everything on Squeaksource: >>>>> http://squeaksource.com/@gUSXOmZmeac-jt63/3ejuPFSA >>>>> In that repo is also the code I use to generate the repository. >>>>> >>>>> Feel free to use the code and the repo but keep in mind that I'm still >>>>> working on it. >>>>> >>>>> Max >>>>> >>>>> >>>>> >>>>> On 16.01.2011, at 11:38, Geert Claes wrote: >>>>> >>>>>> >>>>>> Hi Max, Can you tell us a little bit about pharogenesis? >>>>>> -- >>>>>> View this message in context: >>>>>> http://forum.world.st/pharogenesis-tp3219872p3219893.html >>>>>> Sent from the Pharo Smalltalk mailing list archive at Nabble.com. >>>>>> >>>>> >>>>> >>>> >>>> >>> >>> >> >> > >
