On Feb 11, 2008 8:57 AM, dormando <[EMAIL PROTECTED]> wrote: > Ciaran wrote: > > > > > > On Feb 11, 2008 8:50 AM, dormando <[EMAIL PROTECTED] > > <mailto:[EMAIL PROTECTED]>> wrote: > > > > Ciaran wrote: > > > > If the user would want to use several serialization formats, > > she would > > > > have to use some prefix to encode the info on the used > > format. And it > > > > seems it can't be made compatible with prefix-unaware > clients. > > > > > > Just to note that I'm not ignoring this. I have no > > constructive input > > > past this point :) Have you tried implementing any of this > yet? > > > > > > > > > I've successfully implemented a cross-platform solution between > the > > > Enyim.Memcached C# client and the Java Spy.Net client after these > > > discussions. Using the existing Transcoder framework built into > the > > > Java client and by providing a patch for the C# client, it works > > > beautifully for me, however it isn't a suitable approach for a > wider > > > cross-platform hit, hopefully if Enyim accepts my patch it will be > > > trivial to achieve this between C# + Java, then there just needs > to be > > > some convincing of all client developers that it is a worthwhile > > goal :) > > > > > > (Based upon Tomash' assertion that flags should effectively *only* > be > > > used to mark things as compressed (and|or) serializable, if > > they're not > > > marked as serialized then the stored cache data is treated as a > raw > > > array of bytes, otherwise the transcoder serialisation routine > > inspects > > > the first byte (i.e. the true data flags shifted out of the > memcached > > > flag and into the cached data, which does make sense really) ) > > > > > > - Ciaran > > > > Could you be convinced to start a new wiki page with code examples > (or > > links to them at least) along with an overview of exactly what > you're > > doing? If we want to build standards it needs to be visible in a > place > > where we can start tweaking it and add notes. I don't have working > code, > > so I'm hoping you'll be generous :) > > > > > > What, you mean a bit like [1] ? :) I did that a while back, working > > code's fine and of course I'm happy to drop whatever in to help (its in > > my interest to have memcached work in a nice cross-platform way as thats > > how I'm trying to use it!), but I don't want to drop any code examples > > there just as yet because the requisite patches for the C# client I'm > > currently using haven't been accepted yet, and they may go a different > > way (hopefully not *too* different however!) > > > > Reading the current utf-8-in-perl discussion however it seems that this > > is a directly competing requirement to mine *sigh*, in this approach it > > seems that people are happy to use the flags to encode serialisation > > specific information ? > > > > > [1]http://www.eu.socialtext.net/memcached/index.cgi?cross_platform_serialisation_support > > No, I mean code, regardless of the patched state. "This is how I solved > the problem" - not just the list of flags. Include the path to enyim in > the page too. Mark across the top that this is experimental and for > discussion. > > and yes, I'm asking this because there're some conflicting discussions > about flag usage.
Sure, ok. I'll try to put some code up there tonight, but like I said its not very exciting at the moment, its just a mechanism for how cross-platform serialisation might be sensibly achieved rather than a real-world example (its real-world enough for me as my primary need is to share UTF8 encoded strings stored as byte-arrays between the platforms ;) ) - Ciaran > > > -Dormando > -- - Ciaran
