+1
I'd personally be in favour nuking the whole target, but this works in
a principal of least suprise to people running rake -T
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
Puppet Developers group.
To post to
The main reason I always use tmp directories for my tests is just that
I always know I'm starting from a clean state; I've had too many
problems that were either hidden because I had a complete
configuration or only existed when doing crazy things in an already
set up environment.
Yes,
Oh, one more thing -- this isn't required or anything, but it would be
simpler to read this commit series if you ran it through 'rebase -i
master, and squashed paired commits.
I've added a note on the DevelopmentLifecycle to let people who may
not know about rebase -i they can and should do
2009/2/16 Luke Kanies l...@madstop.com:
Are these changes to ENV done in the same process or in a subprocess
after a fork?
If they're they same process, they should probably clean the value up
to its previous value.
The patch does that by setting @rubylib and restoring it after:
+
I've tried to capture all the key features I've seen in this thread
below to eventually translate into feature requests/stories:
* Namespaces
* Hierachical grouping
* Collapsing information for coarse grained logic (has_interface?)
* External fact loading mechanisms (note I think these depend
I would maybe describe this as 'setting TTL on fact values' or
something similar -- the TTL is the critical bit, since we're already
caching values.
Yeah.
That seems to be it. Let us know when you're done. :)
Will start when I can see 1.9.x branch in say github.com/reductivelabs/facter/
Based on our discussions from our last dev call, I think the branch
should instead be called 'next', rather than '1.anything'.
Indeed.
I'll make and publish such a branch right now.
Thanks I'll start going over the tickets and generating new ones
shortly after (maybe tomorrow)
Paul
2009/2/11 Luke Kanies l...@madstop.com:
There were apparently some circumstances that
resulted in the connection not being closed; this just closes
it every time if it's still open after the rpc call is complete.
+1, completely makes sense to tear down in an ensure block.
Paul
Using the fact value has been completely unreliable. I'll be bug
reporting it once I've worked out the shape of the problem better.
That's a completely fine reason but concerning none the less
$ irb
require 'facter'
= true
Facter.value(operatingsystem)
= Darwin
2009/2/12 Nigel Kersten nig...@google.com:
ahah. So this is the problem. I was testing using ralsh (which I've
become quite insistent on having working properly with my stuff) and
that doesn't load the facts. I need to manually do a Facter.loadfacts
in my provider for that to work.
Another
Blake is mostly looking into whether CouchDb makes more sense than a
relational db, and, if so, how it should be organized (one document
for the whole catalog, one document for each resource, edge, and
catalog, etc.).
Interesting is there anything WIP on github to look at - even if it's
a
(Mostly addressed to Paul Nasrat, as he's started looking at using
Rack in 0.25)
I had an uncomfortable realization about Rack usage by Puppet - we
can't use it for everything without requiring it as a dependency,
which I don't think really works. It's probably acceptable for the
server
2009/1/27 Luke Kanies l...@madstop.com:
Hi all,
We need another dev call, and really should have already done it.
Evening US-time seems to work best, inasmuch as best often still
isn't very good, and Wednesday is best for me. Would tomorrow night
be too soon?
Yeah I've prior commitments
2009/1/21 Thomas Bellman bell...@nsc.liu.se:
Signed-off-by: Thomas Bellman bell...@nsc.liu.se
I may be missing something, but I'm just trying to understand what
problem you're trying to solve by this addition.
The original bug report doesn't really say. Do you mind giving a
practical example
2009/1/26 James Turnbull ja...@lovedthanlost.net:
Signed-off-by: James Turnbull ja...@lovedthanlost.net
Looks good but untested my end atm.
I'm wondering if this isn't a more general symptom of how handle
turning on and off of conditional types and providers being unclear.
Obviously lets fix
There is a more general ability, but it's predicated on provider
features, and files don't use providers. See the 'required features'
in various types in the type reference.
Sure, I guess the language I'd use to describe it is that it is
conditional capabilities of the type(s). Files in
2009/1/25 James Turnbull ja...@lovedthanlost.net:
Signed-off-by: James Turnbull ja...@lovedthanlost.net
---
CHANGELOG |2 ++
lib/facter/network.rb | 11 +++
lib/facter/util/ip.rb | 13 +
3 files changed, 26 insertions(+), 0 deletions(-)
create
There are some tests for util/ip.rb already, we probably should add
some for describing get_network_value.
diff --git a/lib/facter/network.rb b/lib/facter/network.rb
new file mode 100644
index 000..1469902
--- /dev/null
+++ b/lib/facter/network.rb
@@ -0,0 +1,11 @@
+require
Signed-off-by: Paul Nasrat pnas...@googlemail.com
---
lib/facter/{ipmess.rb = interfaces.rb} |2 +-
spec/unit/{ipmess.rb = interfaces.rb} |2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
rename lib/facter/{ipmess.rb = interfaces.rb} (98%)
rename spec/unit/{ipmess.rb
The hardest part for me maintaining these files is knowing what the
behaviour *should* be. David Lutterkort has generally done a good job
of vetting the patches and handling the bugs, but he's been only
sporadically available recently, so we do need more people helping out
with rpm/yum
2009/1/24 James Turnbull ja...@lovedthanlost.net:
Luke Kanies wrote:
Seems like a good idea.
+1
I'll file a ticket and push a branch to github.
Paul
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
Puppet
2009/1/24 Paul Nasrat pnas...@googlemail.com:
2009/1/24 James Turnbull ja...@lovedthanlost.net:
Luke Kanies wrote:
Seems like a good idea.
+1
I'll file a ticket and push a branch to github.
http://projects.reductivelabs.com/issues/show/1888
git pull git://github.com/pnasrat/facter.git
Unfortunately I completely halted what I did on the filebucket over REST
refactoring I did earlier. If it proves to be more important to have
this for 0.25 (on the same side what's the ETA for this version?), then
I'll perform a context switch to start working on the filebucket, but
I'm
2009/1/17 James Turnbull ja...@lovedthanlost.net:
Any chance we can break this out into functions a bit more, I know
it's just the rakefile and helper stuff but it's a long block.
So I'm guessing that we need to do say:
feature/master/win32 which seems a little verbose - is that right?
So I was looking through facter thinking about the problem of
resolving facts based on namespace eg: ipaddress_eth1 ...
I had a quick look at the yaml output as I was thinking about what do
facts mean. Initially my eye was drawn to quoting discrepancies. Then
I started to think ok so if
Any chance we can break this out into functions a bit more, I know
it's just the rakefile and helper stuff but it's a long block.
Sure - I just copied the task from the Puppet Rakefile - because I use
it there.
So I'm guessing that we need to do say:
feature/master/win32 which seems a
A tiered data model works a lot better for some (all?) facts.
Especially if you can traverse up and down the tier for key=value and
structured output.
Yeah I sort of see the easy wins for not breaking stuff in having
either to_s methods and some way of key flattening to map to our
currently
2009/1/15 James Turnbull ja...@lovedthanlost.net:
Signed-off-by: James Turnbull ja...@lovedthanlost.net
+1
Removing circular dependencies can only be a good thing.
Paul
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google
2009/1/4 Luke Kanies l...@madstop.com:
I've made the IPMess stuff a lot less messy, and refactored
a lot of the util/ip module, including naming it more sensibly.
The biggest changes are that I moved the big case statement into
a constant and then used a bit of dispatch-style logic to use
---
lib/puppet/provider/mailalias/aliases.rb | 13 -
1 files changed, 12 insertions(+), 1 deletions(-)
diff --git a/lib/puppet/provider/mailalias/aliases.rb
b/lib/puppet/provider/mailalias/aliases.rb
index 8b5c456..fa3ef9f 100755
---
2009/1/15 Paul Lathrop p...@tertiusfamily.net:
Bug in the aliases provider which causes problems if the recipient field
contains a ':'
+1
Thanks for taking the effort to write a test.
Paul
--~--~-~--~~~---~--~~
You received this message because you are
Signed-off-by: Paul Nasrat pnas...@googlemail.com
---
autotest/puppet_rspec.rb |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/autotest/puppet_rspec.rb b/autotest/puppet_rspec.rb
index c19c294..2bc2fbb 100644
--- a/autotest/puppet_rspec.rb
+++ b/autotest
2009/1/4 Luke Kanies l...@madstop.com:
Signed-off-by: Luke Kanies l...@madstop.com
Sorry to be anal about whitespace but if it is a fixup patch - you
missed one, otherwise all good :)
git show --check
commit d92af476f38d3315cd456c1f7b44eacc62eed578
Author: Luke Kanies l...@madstop.com
Date:
You add some trailing whitespace in this patch, could you trim that and resend.
warning: 9 lines add whitespace errors.
git show --check
commit 73998a3a6e8a46a3bf7b618985621080cb83d1fd
Author: Luke Kanies l...@madstop.com
Date: Fri Jan 2 16:16:21 2009 -0600
Fixing indentation everywhere
Please let me know if you'll be joining, as we'll once again do a
Skype call that I'll organize. I'll also be recording and publishing
the call, as I have done the last two.
I should be able to make this one
Paul
--~--~-~--~~~---~--~~
You received this
I assume everyone else is experiencing the drastic speed problems I'm
having on github.
Anyone think it's time to dump them?
It's been quite frustrating at times.
I've got gitosis set up on redutivelabs, and it's very easy to set up
new repositories (although, obviously) not as easy as
That would definitely be a good thing. Facter also doesn't use facts
with same number of confines in a predictable order (#1291)
Ouch.
For domain name resolution in particular, I see no reason to have all
these independent resolutions with no confines, since we do want them to
be executed
For domain name resolution in particular, I see no reason to have all
these independent resolutions with no confines, since we do want
them to
be executed in a fixed order. I provide a patch in #1654 that merges
all
these resolutions into one.
I'm sure everyone wants them executed in a
Yes.
You're right that it's not in the branch, but a patch was submitted by
Paul Nasrat. And, I think, the ticket was closed. :/
http://projects.reductivelabs.com/issues/show/1642
I'm guessing it's just not been pushed up to github
Paul
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/puppet/type/file/checksum.rb |2 +-
spec/unit/type/file.rb | 26 +-
2 files changed, 26 insertions(+), 2 deletions(-)
diff --git a/lib/puppet/type/file/checksum.rb b/lib/puppet/type/file/checksum.rb
2008/10/8 Luke Kanies [EMAIL PROTECTED]:
On Oct 7, 2008, at 2:35 PM, Martin Englund wrote:
Gang,
how do I force facter to calculate a fact value each time it asked for
instead of at startup? (see facter feature #1424 for the details)
Basically, you just need to have the calculation
Anyway, just throwing that out there.
I'll have a look on the code to see how I can refactor this.
Don't expect too much though :-)
If you want to collaborate talk through ideas on this, I'm happy to
work on this area too.
--~--~-~--~~~---~--~~
You received
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
spec/unit/network/handler/fileserver.rb |8
1 files changed, 8 insertions(+), 0 deletions(-)
diff --git a/spec/unit/network/handler/fileserver.rb
b/spec/unit/network/handler/fileserver.rb
index cce00d3..4ba8e71 100644
--- a/spec
an empty response instead of a file response.
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/puppet/network/handler/fileserver.rb | 10 +++---
1 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/lib/puppet/network/handler/fileserver.rb
b/lib/puppet/network/handler
.
The change to the new API meant that we lost the existance test of plugins
mount directory. It was failing when the client was mounting module plugins that
weren't existing (like facter and no custom facts defined).
Signed-off-by: Brice Figureau [EMAIL PROTECTED]
Signed-off-by: Paul Nasrat [EMAIL
Thanks to Brice for picking up on this, I've checked through this and added a
missing test.
Apologies for introducing it in the first place :(
b/lib/puppet/network/handler/fileserver.rb | 13 +
b/spec/unit/network/handler/fileserver.rb | 39 -
*) Since the refactoring of FileServer, the plugin mount were broken:
+ reclist() signature was modified, but not the PluginMount call site
+ then list() was requiring 'file_serving/fileset', but since PluginMount
was using reclist directly, it wasn't requiring fileset.
+ the
ge commands.
First, they tend to be slow -- slow enough that you don't really want
them running every time you run your unit tests. This means you'd
probably want to split out the rake tasks so that we default to only
running the unit tests, and we have to go out of our way to run the
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/puppet/network/handler/fileserver.rb |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/lib/puppet/network/handler/fileserver.rb
b/lib/puppet/network/handler/fileserver.rb
index 3e62cdb..89b0d53 100755
--- a/lib/puppet
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/puppet/file_serving/metadata.rb |6 +++---
spec/unit/file_serving/metadata.rb | 18 +-
2 files changed, 20 insertions(+), 4 deletions(-)
diff --git a/lib/puppet/file_serving/metadata.rb
b/lib/puppet/file_serving
On 25 Sep 2008, at 18:14, Luke Kanies wrote:
Is it worth just switching the existing fileserver to use stuff in
file_serving for everything, and backport whatever changes we need to?
Yeah that's why I started by writing tests for the existing api so we
could surround it and then just swap
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
spec/unit/network/handler/fileserver.rb | 133 +++
1 files changed, 133 insertions(+), 0 deletions(-)
create mode 100644 spec/unit/network/handler/fileserver.rb
diff --git a/spec/unit/network/handler/fileserver.rb
I've taken a different approach (working with my colleague Dan Moore) as
suggested by Luke.
Firstly we wrote the tests to surround the current implementation of
FileServer::Mount to capture the current API.
Then we changed the fileserver implementation to use fileset internally to
perform the
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/puppet/file_serving/metadata.rb |6 +++---
spec/unit/file_serving/metadata.rb | 18 +-
2 files changed, 20 insertions(+), 4 deletions(-)
diff --git a/lib/puppet/file_serving/metadata.rb
b/lib/puppet/file_serving
On 19 Sep 2008, at 16:59, Luke Kanies wrote:
Can you compare this to the current state of the class in the master
branch?
Sure.
Looking at the code it looks as the issue fixed in this patch is
present there too:
master is at:
5b9dd01326a61b9ae89ae978e29a8170f76deb5e Merge commit
On 19 Sep 2008, at 17:00, Luke Kanies wrote:
As mentioned on the ticket, I think the right thing for now is to just
throw a warning if someone tries to use anything other than md5 when
remote file copying, since this change would introduce a backward
compatibility issue.
It will all
On 18 Sep 2008, at 22:16, Jos Backus wrote:
On Thu, Sep 18, 2008 at 10:05:48PM +0100, Paul Nasrat wrote:
I've been seeing it up and down today also. You might try and hit it
again shortly after. note if you're updating an issue, I've often
found it updates but I still get the 500 so don't
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
spec/unit/type/file.rb |6 +-
1 files changed, 1 insertions(+), 5 deletions(-)
diff --git a/spec/unit/type/file.rb b/spec/unit/type/file.rb
index 9fb1900..3e94bc1 100755
--- a/spec/unit/type/file.rb
+++ b/spec/unit/type/file.rb
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
spec/unit/type/file.rb | 26 ++
1 files changed, 22 insertions(+), 4 deletions(-)
diff --git a/spec/unit/type/file.rb b/spec/unit/type/file.rb
index 675cbe9..9fb1900 100755
--- a/spec/unit/type/file.rb
+++ b/spec/unit
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/puppet/network/handler/fileserver.rb | 25 ++---
spec/unit/type/file.rb | 17 ++---
2 files changed, 24 insertions(+), 18 deletions(-)
diff --git a/lib/puppet/network/handler
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
spec/unit/type/file.rb | 18 ++
1 files changed, 18 insertions(+), 0 deletions(-)
diff --git a/spec/unit/type/file.rb b/spec/unit/type/file.rb
index 7f9688f..26a8ff9 100755
--- a/spec/unit/type/file.rb
+++ b/spec/unit/type
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/facter/kernelrelease.rb | 15 +++
1 files changed, 15 insertions(+), 0 deletions(-)
diff --git a/lib/facter/kernelrelease.rb b/lib/facter/kernelrelease.rb
index 0f8a643..bff7a90 100644
--- a/lib/facter/kernelrelease.rb
+++ b
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/facter/macaddress.rb | 14 ++
1 files changed, 14 insertions(+), 0 deletions(-)
diff --git a/lib/facter/macaddress.rb b/lib/facter/macaddress.rb
index 30d4b59..98e38b8 100644
--- a/lib/facter/macaddress.rb
+++ b/lib/facter
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/facter/kernel.rb |8 +++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/lib/facter/kernel.rb b/lib/facter/kernel.rb
index 01fe85b..5a0eb19 100644
--- a/lib/facter/kernel.rb
+++ b/lib/facter/kernel.rb
@@ -1,3 +1,9
Signed-off-by: Paul Nasrat [EMAIL PROTECTED]
---
lib/facter/domain.rb | 14 ++
1 files changed, 14 insertions(+), 0 deletions(-)
diff --git a/lib/facter/domain.rb b/lib/facter/domain.rb
index 724dd1b..57b04f6 100644
--- a/lib/facter/domain.rb
+++ b/lib/facter/domain.rb
@@ -62,3
On 17 Jul 2008, at 13:58, Paul Nasrat wrote:
The following patches to list enable very simple windows support for
facter, in order to enable eventual puppet windows support.
The idea is to get some feedback/review of the approaches I've taken
and try figure out what needs to be done.
One
201 - 266 of 266 matches
Mail list logo
