~themue/juju-core/027-http-synctools : changes

~themue/juju-core/027-http-synctools » Changes from revision 1283

From Revision 1283 to 1264

Rev	Summary	Authors	Date
1283	synctools: merged trunk synctools: merged trunk	Frank Mueller	10 years ago
1282	synctools: merged trunk synctools: merged trunk	Frank Mueller	10 years ago
1281	synctools: changed to use http access to juju-dist synctools: changed to use http access to juju-dist	Frank Mueller	10 years ago
1280	[r=jameinel] cleaner: implemented the cleaner worker [r=jameinel] cleaner: implemented the cleaner worker The cleaner worker listens to events raised by the cleaner watcher. This signals that there are documents marked for removal. So the worker calls state.Cleanup() to remove them. https://codereview.appspot.com/10148045/	Frank Mueller	10 years ago
1279	[r=jameinel] cmd/juju: Make a test deterministic [r=jameinel] cmd/juju: Make a test deterministic TestGatherDescriptionsInParallel was timing dependent, and occasionally failed on my system. Rather than just increasing the timeout arbitrarily, I changed the code to be more 'event' based. Each plugin requires another plugin to have started before that plugin will return. Thus, if we don't start the plugins in parallel, the plugin will deadlock. Because the failure mode is 'deadlock', I wrapped the call in a select timeout. This means that if it does fail, we have a garbage goroutine and process lying around (but it should never fail). On the plus side, this makes it actually run faster, since each subprocess doesn't waste time sleeping just so we can measure it. (from 350ms down to normally around 15ms.) This is the only test that was failing (based on occasional timing hiccups) for me, which allows us to have this in tarmac.	John Arbash Meinel	10 years ago
1278	Update the validMachine regex for containers Update the validMachine regex for containers Yay regex. R=wallyworld, fwereade CC= https://codereview.appspot.com/10215043	Tim Penhey	10 years ago
1277	deploy refactoring deploy refactoring GUI deploy and CLI deploy are different enough that the common statecmd caused more problems than it solved. Testing is noticeably improved. juju.Conn.DeployService and juju.Conn.AddUnits are now probably ready to move to some other place that just requires a state connection (and not an environment as well); juju.Conn.PutCharm needs some love too, and thought devoted to how we're going to put local charms over the API. But, for now, the various bits all happen in the right place (except that the CLI once again downloads store charms and uploads them itself, rather than taking advantage of that functionality on the server side. This can and will be fixed, but not this CL). R=rog, jameinel, mue CC= https://codereview.appspot.com/10166044	William Reade	10 years ago
1276	state: Service ConfigSettings methods state: Service ConfigSettings methods Config, SetConfig, and SetConfigYAML methods have been dropped in favour of ConfigSettings and UpdateConfigSettings, which use sensible types. Clients are expected to parse their own damn data and supply a sensible format. R=mue, rog CC= https://codereview.appspot.com/10083047	William Reade	10 years ago
1275	state: Service.SetConfigYAML state: Service.SetConfigYAML ...now works as intended. Don't get too attached to it though. R=thumper, wallyworld, rog CC= https://codereview.appspot.com/10172045	William Reade	10 years ago
1274	state: Unit.WatchConfigSettings state: Unit.WatchConfigSettings state.Settings is bad, because it allows any client the ability to read/change the associated document without any sort of validation; and sending them down a channel is independently bad because the documents don't contain the right data for the unit context. So, I just made it an EntityWatcher instead. Works just as well, is rather simpler, doesn't expose massive potential bugs. Win! R=thumper, wallyworld, rog CC= https://codereview.appspot.com/10180043	William Reade	10 years ago
1273	Add ParentId() to Machine Add ParentId() to Machine This branch provides the ability to ask a container for its parent machine's id. R=thumper, fwereade CC= https://codereview.appspot.com/10203043	Ian Booth	10 years ago
1272	state: Unit.ConfigSettings state: Unit.ConfigSettings Better name, better type; suitable changes propagated amongst clients. R=thumper, wallyworld, rog CC= https://codereview.appspot.com/10169044	William Reade	10 years ago
1271	charm: trivial error consistency tweak charm: trivial error consistency tweak expect followup changing charm package to use errors.NotFoundError R=thumper, wallyworld, rog CC= https://codereview.appspot.com/10178043	William Reade	10 years ago
1270	Fix machine tag for containers Fix machine tag for containers Container ids can't be used in file names since they contain "/". Replace the "/" with a "-". R=thumper CC= https://codereview.appspot.com/10019049	Ian Booth	10 years ago
1269	Update add machine command to create containers Update add machine command to create containers The add machine command accepts additional command line arguments to specify a container to create. The container can be on a new machine or an existing instance. Container constraints are also supported eg create a container with 2G memory. The state is updated with the new machine document but the machine watcher ignores the machine since there's no code to provision containers yet. examples: juju add-machine /lxc juju add-machine --constraints mem=4G /lxc juju add-machine 1/lxc R= CC= https://codereview.appspot.com/9886045	Ian Booth	10 years ago
1268	Fix compile errors in the Azure provider skeleton. Fix compile errors in the Azure provider skeleton. I broke the build by omitting some imports and such. I forgot that I hadn't built my branch yet, and since lbox usually complains about some build problem or other that's unrelated to my changes, I thought it would run a build either when proposing or submitting. There was talk about enforcing a clean build at that state a while ago. Turns out it never did. R=gz, jeroen.vermeulen CC= https://codereview.appspot.com/10184043	Jeroen T. Vermeulen	10 years ago
1267	charm: rewrite Config type, add Settings type charm: rewrite Config type, add Settings type The existing Validate method was extremely unwieldy; in the course of decomposing it, and adding parsing for a different input format, the entire module changed in almost every respect. Hopefully the change demonstrates its value in this CL alone, what with the changes to the previous clients of Validate; there's only one large- scale behaviour change, which is that charm upgrades no longer error on config changes of incompatible type: they now just silently reset the values (as agreed with m_3 in oakland -- not a big deal, just a slight improvement in consistency). R=thumper, wallyworld, rog CC= https://codereview.appspot.com/10162045	William Reade	10 years ago
1266	Rollback r1264 which broke the build Rollback r1264 which broke the build The Azure provider skeleton does not build, so revert it for now. R=rog CC= https://codereview.appspot.com/9714046	Martin Packman	10 years ago
1265	Skeleton for prospective Azure provider. Skeleton for prospective Azure provider. Take this with a grain of salt. We won't mind changing it at all. It's just an outline of what we'll need to implement in order to build an Azure provider, as a starting point to get the work going. R=fwereade, Danilo CC= https://codereview.appspot.com/10046045	Jeroen T. Vermeulen	10 years ago
1264	Add install-dependencies target to Makefile. Add install-dependencies target to Makefile. This makes it slightly less work to get a juju-core branch ready for development. Just "make inst<tab><enter>" will take care of most of the dependencies for you. Not much use for those who already have their machines set up for juju development, but it takes away some of that crucial initial friction for those who don't. R=julian.edwards, thumper, jameinel, jeroen.vermeulen CC= https://codereview.appspot.com/10046044	Jeroen T. Vermeulen	10 years ago

Older »