Closed (fixed)
Project:
Flexinode
Version:
master
Component:
Code
Priority:
Critical
Category:
Bug report
Assigned:
Unassigned
Reporter:
Created:
29 May 2006 at 14:44 UTC
Updated:
13 Jul 2006 at 06:45 UTC
Jump to comment: Most recent file
Comments
Comment #1
morbus iffWhoops. Small fix.
Comment #2
morbus iffCommitted.
Comment #3
scott.mclewin commentedI have a simple question about this change. I use flexinode on my 4.7 based site. I've been tracking HEAD rather than the 4.7 branch (due in part to it not appearing to be official yet). In reading through this issue and the long running conversation it links to regarding changes to the links system, I didn't get a clear answer on whether or not this change will work in a 4.7 site or whether it requires core from HEAD.
Is this change a point at which implicit 4.7 support from HEAD stops? (I don't know that it is, but I also don't know that it is not). I'm just looking for clarity. It seems that it is time to track flexinode from the 4.7 tag rather than head.
Thanks
Comment #4
morbus iffLemme get this straight before I respond:
* You're using Drupal 4.7 on your site.
* You're using the Flexinode module from HEAD.
If the above is true, then the Flexinode module in HEAD will malfunction if you try to run it on your Drupal 4.7 site. This is purely expected, as you're running mismatched versions (generally speaking, you should be running 4.7 with 4.7 stuff, or HEAD stuff with HEAD stuff, not a mish-mash). The only Flexinode builds that are intended to work with Drupal 4.7 are those branched to DRUPAL-4-7. HEAD versions of modules, if they're being regularly maintained, may or may not work with release versions of Drupal (it really depends if the developers are making new features to an existing module or merely keeping the code working with the latest in-development Drupal code). For Flexinode's case, it's keeping up with the latest code and API from Drupal HEAD. Nowadays, Flexinode is pretty much in a maintained state only - very few "features" will be added as CCK is largely the way forward.
With that said, Berkes is the maintainer for the Flexinode branch of 4.7 -- when he feels enough bug fixes and tweaks have been made to warrant a new release intended for 4.7, he will go ahead and make one.
Comment #5
scott.mclewin commentedYou have it strait. I've been working against 4.7 since it was head and pre-beta. As modules maintainers introduced explicit 4.7 branches I made the call on a module by module and then a change by change basis to move from tracking the head version to tracking the 4.7 version. As of this change, flexinode is at that point. I recognize that I have a version mismatch - I study the changes as they come out for the shrinking subset of modules that I draw out of head for my 4.7 based site. FWIW the site is not yet in production - I'm still working on some custom logic to make 95% of the displays on the site filter the content to nodes that are physically close to the user's configured location. That involves submitting patches, which are often faster to get turned around in head. I'm hedging in a few cases where there is no 4.7 branch that by the time I am ready for production that the module maintainer will have introduced explicit 4.7 support.
You've done a good job of gently describing why what I'm doing is a bad idea (or perhaps just a risky idea). I recognize that. It is a risk I've choosen to take.
Up until this change flexinode didn't have any obvious (to me) changes that made it not work in 4.7 and it has indeed worked well. I realize that my question raised "guy doing stupid things and now asking for support" alarm bells. I hope I've put the right context behind the question in this reply, and will pose the simple question again:
I believe that the links support changes from this patch will in no way work in 4.7 because they rely on the link theming capabilities being introduced in 4.8. Is that correct?
I'm asking because I'm either going to adopt this change too, and test it in an environment that I know you won't (and should not be) testing it. Or I'm going to switch to using the 4.7 tagged flexinode. Switching to the 4.7 means I'll studying it vs. head to see what is 'missing', test that, and perhaps submit requests against flexinode to bring some changes across to 4.7. I believe the latter is the path I'll be taking. I'm simply looking for a sanity check from somebody who I know is more of a Drupal expert than I am.
I do appreciate your quick and thorough response. Thank you for that. I get frustrated by unpunctuated one sentence answers I sometimes get in response to two paragraph questions.
Comment #6
morbus iffyestherecentchangeisfor48onlyn00b
How's that for an unpunctuated one sentence answer? ;) Note, naturally, that this change IS quite small and only affects one file. One of the things you could consider doing is making a patch that /removes/ this change, and applying that patch everytime you update to the latest HEAD. That way you're still using the latest code, but then are repatching it (in effect, backporting it) to the Drupal version you're using.
Comment #7
RayZ commented@Scott: Given the current limitations of project versioning for contrib and the varying choices by contrib developers of whether to track HEAD of drupal core, I think your evaluation of which branch to track on a "module by module and then a change by change basis" is the only one I've found to be workable, especially when using newer modules on a site that's not yet in production.
This is a good example of why the effort toward a new method for contributed project releases and versions is so important. See also http://groups.drupal.org/node/847.
Comment #8
scott.mclewin commentedMorbus: I think the n00b part put me over the top. I'll put up with unpunctuated single line answers when they make me laugh.
RayZ: I was unaware of those discussions. I probably should not have sat here reading them so late (in my part of the world it is approaching 2am) after a seven hour rehearsal and six hours before I head back to work, but I'm up and interested. I'm glad attention is going into better version management for modules. I've been frustrated at my lack of versioning flexiblity for the one module I maintain (wishlist). I do appreciate the validation that my hackish method of tracking changes is unfortunately the best option available to me.
Ok, I'm done accidentally hijacking this issue.
Comment #9
(not verified) commented