I should have added, I think the work so far should continue to live on its branch because until integrating it with the rest of the client (which is the more interesting bit to the puzzle) really gets underway the approach is still really unproven and there is little benefit to it being on trunk before that happens.
Robbie On 6 April 2012 12:31, Robbie Gemmell <[email protected]> wrote: > Option 1 seems like the only sensible course of action, options 2 and > 3 would just be continuing further down the paths that lead us where > we are now. At the end of this process there should really be less > arbitrary segregation of codepaths and less legacy cruft lying around, > not more of it. > > Robbie > > On 5 April 2012 16:34, Rajith Attapattu <[email protected]> wrote: >> Once we get agreement on the destination work, the next step is to >> integrate it into the code base. >> I see the following options. Please feel free to provide your feedback on >> the options provide or even suggest new options. >> I'm quite happy to work with any option the majority feels the best. >> >> 1. Integrate into both 0-8 and 0-10 code paths. >> >> 2. Integrate into 0-10 only. - This would be tricky as there is a fair bit >> of common code that uses destination stuff. >> >> 3. Have a new implementation of Session, Producer, Consumer etc for 0-10 >> that exclusively uses the new destination impl. >> We could then use AMQConnectionDelegate_0_10 as a point to plug in this >> new structure when 0-10 is used. >> >> (Note: I'm not advocating a rewrite, rather to reuse as much code as >> possible, and by moving necessary pieces from common code into the specific >> implementation we could do #2 without being encumbered by common code). >> The flip side is code duplication. >> >> Regards, >> >> Rajith --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
