Hi Anil, Inlined below.
On Sat, May 3, 2014 at 8:19 AM, anil gupta <anilgupt...@gmail.com> wrote: > @Gabriel, It would be really nice if we dont need to copy anything in lib > directory in Squirell. Would you mind explaining high level how you are > achieving this? What all input parameters user will need to provide while > connection to HBase through Squirell? Just to be clear, the Phoenix jar is still needed, it's just that it can be put on any location on the machine and supplied to SQuirreL via the "Additional classpath" setting for the driver. The thing in Phoenix that prevents this from working in general is the dynamic loading of configuration files and (at least in Phoenix 3.0) the loading of query endpoint return value classes via the current thread's context classloader. This is a problem with SQL tools that load classes dynamically and do things with JDBC connections in threads where a special context classloader is set up. The main change in the patch on PHOENIX-966 is just ensuring that the current thread's context classloader is set up properly on calls where dynamic loading of Phoenix or HBase resources is done. > > Also, what are our plans for trying better looking DB Browser tools like > AquaData and DBVisualizer. Latest Aquadata already supports Cassandra, Hive > and Mongo. Aquadata and DBVisualizer are more popular DB browsers. IMHO, > Squirell UI is not as good as UI of AquaData and DBVisualizer. If we can > make Phoenix work with these two applications also then it will be a huge > win for us. > I'm currently working on making Phoenix work properly with another commercial tool from a large vendor, but I think that the general principles are the same as getting it to work with SQuirreL, as long as we're talking about JDBC functionality. I actually just tried my current work in progress from PHOENIX-966 with DBVisualizer and it appears to work fine. I'll give it a try with Aquadata as well. - Gabriel > > On Fri, May 2, 2014 at 9:49 PM, Gabriel Reid <gabriel.r...@gmail.com> wrote: > >> Hi guys, >> >> Closely related to this, I just opened PHOENIX-966, which involves >> making it possible to use Phoenix without having to copy anything to >> the lib directory in SQuirreL. The patch in that ticket will change >> the behavior needed for setting up Phoenix with SQuirreL and other >> JDBC-based SQL tools. This is no reason to not update the docs with >> information on copying the config as you suggested, but I just wanted >> to give you a heads-up about it. >> >> - Gabriel >> >> >> On Sat, May 3, 2014 at 3:25 AM, James Taylor <giacomotay...@gmail.com> >> wrote: >> > I agree. Would you be up for sending us a patch with the updates? You >> could >> > enhance our directions to include what you need to do to connect to a >> > secure cluster as well. The website is mostly done through markdown and >> > accessible through svn. See phoenix/build.txt. >> > >> > The download file, for example is in <top level site >> > dir>/site/source/src/site/markdown/download.md. >> > >> > Thanks, >> > James >> > >> > >> > On Fri, May 2, 2014 at 6:20 PM, anil gupta <anilgupt...@gmail.com> >> wrote: >> > >> >> Hi James, >> >> >> >> http://phoenix.incubator.apache.org/download.html >> >> >> >> Instruction to use Phoenix with Squirell does not mention conf file to >> be >> >> copied in classpath of Squirell. I think we should mention a step for >> >> copying the hbase conf files in classpath of Squirell. Also, it would be >> >> good if you provide an example of a connection string using a ZK quorum >> of >> >> distributed cluster rather than "localhost". >> >> >> >> This is just my feedback. >> >> >> >> -- >> >> Thanks & Regards, >> >> Anil Gupta >> >> >> > > > > -- > Thanks & Regards, > Anil Gupta
