On 12/22/2015 11:45 PM, William Bell wrote: > Why would someone stay on 5.3.x instead of upgrading to 5.4? Why backport > when you can just upgrade?
We just had this same discussion on the dev list. Anshum wants to cut a 5.3.2 release, somebody asked the same thing you did, and this is what I wrote in reply: ======== I am using a third-party custom Solr plugin. The latest version of that plugin (which I have on my dev server) has only been certified to work with Solr 5.3.x. There's a chance that it won't work with 5.4, so I cannot use that version yet. If I happen to need any of the fixes that are being backported, an official 5.3.2 release would allow me to use official binaries, which will make my managers much more comfortable than a version that I compile myself. Additionally, the IT change policies in place for many businesses require a huge amount of QA work for software upgrades, but those policies may be relaxed for hotfixes and upgrades that are *only* bugfixes. For users operating under those policies, a bugfix release will allow them to fix bugs immediately, rather than spend several weeks validating a new minor release. There is a huge amount of interest in the new security features in 5.3.x, functionality that has a number of critical problems. Lots of users who need those features have already deployed 5.3.1. Many of the critical problems are fixed in 5.4, and these are the fixes that Anshum wants to make available in 5.3.2. If a user is in either of the situations that I outlined above, upgrading to 5.4 may be unrealistic. ======== I am already using a homegrown version (5.3.2-SNAPSHOT) for my dev server, because I require a fix that's only in 5.4: SOLR-6188. Unless the release manager objects, I plan to add this fix to 5.3.2. Thanks, Shawn