Re: Deploying a new 2.0.7-SNAPSHOT
Daniel Kulp wrote: On May 27, 2008, at 12:26 PM, Eoghan Glynn wrote: Thanks Dan, Following the instructions on [1] I've run into an authentication failure after only a few modules are deployed to the repo. Is there a way of rolling back the partially deployed artifacts, do you know? Just redeploy. Not a big deal. For the most part, follow the "HIGHLY RECOMMENDED" section on that page to get the master mode stuff working. By default, maven creates 3 connections for each file being deployed which is: 1) slow as it's got to authenticate each time 2) unreliable - that many connections eats the entropy on both sides very quickly and may be the source of the authentication issues. Makes sense. BTW what's the deal on getting the PKI set up, to avoid keyboard interactive auth? Cheers, Eoghan Another option is to change the protocol used from scpexe to just scp (top level pom, just change in you working copy, deploy, and then reset back). On some OS's, I've had better luck with scp instead of scpexe. Windows works best with scpexe. Linux without master mode works best with scp. I've only done scpexe+master mode on OSX so I'm not sure there. However, master mode with scpexe is the most reliable. Dan Cheers, Eoghan [1] http://cxf.apache.org/release-management.html Daniel Kulp wrote: Eoghan, It's pretty much "ad hoc" when someone needs one or if one of us has a fix in place that they want a user to test out. If you need one, you should feel free to deploy one. That said, I'll try and get a new one out later tonight if you don't get to it first. Dan On May 27, 2008, at 5:54 AM, Eoghan Glynn wrote: Hi folks, What would I need to do to arrange for an up-to-date 2.0.7-SNAPSHOT to be be deployed to the maven repo? Is this normally done regularly or on an ad-hoc basis? The timestamps on the 2.0.7-SNAPSHOT dir[1] in the maven repo suggest approximately once a week, but with a gap last week. Cheers, Eoghan [1] http://people.apache.org/repo/m2-snapshot-repository/org/apache/cxf/cxf/2.0.7-SNAPSHOT/ IONA Technologies PLC (registered in Ireland) Registered Number: 171387 Registered Address: The IONA Building, Shelbourne Road, Dublin 4, Ireland --- Daniel Kulp [EMAIL PROTECTED] http://www.dankulp.com/blog IONA Technologies PLC (registered in Ireland) Registered Number: 171387 Registered Address: The IONA Building, Shelbourne Road, Dublin 4, Ireland --- Daniel Kulp [EMAIL PROTECTED] http://www.dankulp.com/blog IONA Technologies PLC (registered in Ireland) Registered Number: 171387 Registered Address: The IONA Building, Shelbourne Road, Dublin 4, Ireland
Re: Deploying a new 2.0.7-SNAPSHOT
PKI already works. Just put your ssh key into your people.apache.org .ssh dir. On Tue, May 27, 2008 at 2:56 PM, Eoghan Glynn <[EMAIL PROTECTED]> wrote: > Daniel Kulp wrote: > >> >> On May 27, 2008, at 12:26 PM, Eoghan Glynn wrote: >> >> >>> Thanks Dan, >>> >>> Following the instructions on [1] I've run into an authentication failure >>> after only a few modules are deployed to the repo. >>> >>> Is there a way of rolling back the partially deployed artifacts, do you >>> know? >>> >> >> Just redeploy. Not a big deal. >> >> For the most part, follow the "HIGHLY RECOMMENDED" section on that page to >> get the master mode stuff working. By default, maven creates 3 connections >> for each file being deployed which is: >> 1) slow as it's got to authenticate each time >> 2) unreliable - that many connections eats the entropy on both sides very >> quickly and may be the source of the authentication issues. >> > > Makes sense. > > BTW what's the deal on getting the PKI set up, to avoid keyboard > interactive auth? > > Cheers, > Eoghan > > Another option is to change the protocol used from scpexe to just scp (top >> level pom, just change in you working copy, deploy, and then reset back). >> On some OS's, I've had better luck with scp instead of scpexe. Windows >> works best with scpexe. Linux without master mode works best with scp. >> I've only done scpexe+master mode on OSX so I'm not sure there. However, >> master mode with scpexe is the most reliable. >> >> Dan >> >> >> >>> >>> Cheers, >>> Eoghan >>> >>> [1] http://cxf.apache.org/release-management.html >>> >>> Daniel Kulp wrote: >>> Eoghan, It's pretty much "ad hoc" when someone needs one or if one of us has a fix in place that they want a user to test out. If you need one, you should feel free to deploy one. That said, I'll try and get a new one out later tonight if you don't get to it first. Dan On May 27, 2008, at 5:54 AM, Eoghan Glynn wrote: > > Hi folks, > > What would I need to do to arrange for an up-to-date 2.0.7-SNAPSHOT to > be be deployed to the maven repo? > > Is this normally done regularly or on an ad-hoc basis? > > The timestamps on the 2.0.7-SNAPSHOT dir[1] in the maven repo suggest > approximately once a week, but with a gap last week. > > Cheers, > Eoghan > > [1] > http://people.apache.org/repo/m2-snapshot-repository/org/apache/cxf/cxf/2.0.7-SNAPSHOT/ > > > IONA Technologies PLC (registered in Ireland) > Registered Number: 171387 > Registered Address: The IONA Building, Shelbourne Road, Dublin 4, > Ireland > --- Daniel Kulp [EMAIL PROTECTED] http://www.dankulp.com/blog >>> >>> >>> IONA Technologies PLC (registered in Ireland) >>> Registered Number: 171387 >>> Registered Address: The IONA Building, Shelbourne Road, Dublin 4, Ireland >>> >> >> --- >> Daniel Kulp >> [EMAIL PROTECTED] >> http://www.dankulp.com/blog >> >> >> >> > > IONA Technologies PLC (registered in Ireland) > Registered Number: 171387 > Registered Address: The IONA Building, Shelbourne Road, Dublin 4, Ireland >
Re: Deploying a new 2.0.7-SNAPSHOT
Benson Margulies wrote: PKI already works. Just put your ssh key into your people.apache.org .ssh dir. Thanks Benson ... I was sure I'd copied over my public key ages ago, but just checked my home dir on people.apache.org and I didn't even have a ~/.ssh dir, never mind an authorized_keys file. Maybe my apache home dir was blown away at some point, or more likely I was thinking of some other host. Either way, all working now :) Cheers, Eoghan IONA Technologies PLC (registered in Ireland) Registered Number: 171387 Registered Address: The IONA Building, Shelbourne Road, Dublin 4, Ireland
Re: Invalid WSDLs
On May 26, 2008, at 5:32 PM, Benson Margulies wrote:
Yes. I'm also very curious as to whether Dan can shed light as to the
intention: should the CXF wsdl validator *only* worry about CXF-
specific
issues, leaving general validation to generic tools, or does it
ambit to
cover all of the ground?
This is kind of a loaded question. I guess the answer is that it
should catch and wsdl issues that CXF users care about. Seriously,
it currently tackles validation for many of the "common" issues that
users have hit. That includes issues hit by some of IONA's customers
since they are CXF "users" as well. There are a bunch of checks for
WSI-BP compliance as well since CXF SHOULD be generating (and
consuming) WSI-BP compliant wsdl whenever possible.
I'd be careful about changing too many wsdl's. Many wsdl's in the
CXF tree are actually there to test some of these "bad" cases. For
example, the hello_world_xml_bare.wsdl issue you mentionedThat
definitely is an issue in many cases. Currently, if the first parts
match, CXF cannot dispatch it correctly. However, if the soap-
actions are unique, it CAN. There is also a bug logged about it: https://issues.apache.org/jira/browse/CXF-1073
There may be tests related to it. I'm not really sure.
Dan
On Mon, May 26, 2008 at 5:14 PM, Glen Mazza <[EMAIL PROTECTED]>
wrote:
For specific validity questions, you might want to run against
Metro and
see what it says. Although not perfectly reliable, similiar error
messages (or lack thereof) to CXF's might help determine what is not
allowed with WSDL in general compared to what just CXF doesn't like.
Glen
2008-05-26 Benson Margulies wrote:
However, now we get the question of what me mean by 'validate' in
our
tool.
Our validator rejects the wsdl we call hello_world_xml_bare.wsdl
with
the below error, because one operation has a single input part of
element type 'x', and another operation has three inputs parts, the
first of which is also 'x'.
Is this, in fact, unacceptable? Should I change the wsdl to make the
elements distinct? Or is the validator wrong?
On Mon, May 26, 2008 at 4:39 PM, Benson Margulies <[EMAIL PROTECTED]
>
wrote:
What I'm doing is reflecting the existing command line into maven.
On Mon, May 26, 2008 at 3:54 PM, Glen Mazza
<[EMAIL PROTECTED]>
wrote:
I believe we have something like this already on the command-line
(-validate option[1]). Perhaps it would be better to build on
this,
so
command-line/Maven/Ant users can use it as well.
Glen
[1] http://cwiki.apache.org/CXF20DOC/wsdl-to-java.html
2008-05-26 Benson Margulies wrote:
Watchers of checkins can see that I'm inventing a WSDL validation
maven plugin.
First catch:
INFO: Resolve schema http://schemas.xmlsoap.org/wsdl/2003-02-11.xsd
from baseURI:
jar:file:/home/benson/.m2/repository/org/apache/cxf/cxf-common-
schemas/2.1.1-SNAPSHOT/cxf-common-schemas-2.1.1-SNAPSHOT.jar!/
schemas/wsdl/http.xsd,
namespace: http://schemas.xmlsoap.org/wsdl/
[INFO]
[ERROR] BUILD ERROR
[INFO]
[INFO] Non unique body parts, operation [ greetMe ] and
operation [
testTriPart ] have the same body block
{http://apache.org/hello_world_xml_http/bare/types}requestType
---
Daniel Kulp
[EMAIL PROTECTED]
http://www.dankulp.com/blog
Re: Invalid WSDLs
My goal here is to catch inadvertently busted WSDL files when we check them
in. So, using our validator, I'd at least have to exclude some cases that
are supposed to fail.
On Tue, May 27, 2008 at 5:16 PM, Daniel Kulp <[EMAIL PROTECTED]> wrote:
>
> On May 26, 2008, at 5:32 PM, Benson Margulies wrote:
>
> Yes. I'm also very curious as to whether Dan can shed light as to the
>> intention: should the CXF wsdl validator *only* worry about CXF-specific
>> issues, leaving general validation to generic tools, or does it ambit to
>> cover all of the ground?
>>
>
> This is kind of a loaded question. I guess the answer is that it should
> catch and wsdl issues that CXF users care about. Seriously, it currently
> tackles validation for many of the "common" issues that users have hit.
> That includes issues hit by some of IONA's customers since they are CXF
> "users" as well. There are a bunch of checks for WSI-BP compliance as well
> since CXF SHOULD be generating (and consuming) WSI-BP compliant wsdl
> whenever possible.
>
> I'd be careful about changing too many wsdl's. Many wsdl's in the CXF
> tree are actually there to test some of these "bad" cases. For example,
> the hello_world_xml_bare.wsdl issue you mentionedThat definitely is
> an issue in many cases. Currently, if the first parts match, CXF cannot
> dispatch it correctly. However, if the soap-actions are unique, it CAN.
> There is also a bug logged about it:
> https://issues.apache.org/jira/browse/CXF-1073
> There may be tests related to it. I'm not really sure.
>
>
> Dan
>
>
>
>
>
>>
>> On Mon, May 26, 2008 at 5:14 PM, Glen Mazza <[EMAIL PROTECTED]>
>> wrote:
>>
>>> For specific validity questions, you might want to run against Metro and
>>> see what it says. Although not perfectly reliable, similiar error
>>> messages (or lack thereof) to CXF's might help determine what is not
>>> allowed with WSDL in general compared to what just CXF doesn't like.
>>>
>>> Glen
>>>
>>>
>>> 2008-05-26 Benson Margulies wrote:
>>>
However, now we get the question of what me mean by 'validate' in our
>>> tool.
>>
>>>
Our validator rejects the wsdl we call hello_world_xml_bare.wsdl with
the below error, because one operation has a single input part of
element type 'x', and another operation has three inputs parts, the
first of which is also 'x'.
Is this, in fact, unacceptable? Should I change the wsdl to make the
elements distinct? Or is the validator wrong?
On Mon, May 26, 2008 at 4:39 PM, Benson Margulies <
[EMAIL PROTECTED]>
>>> wrote:
>>
>>> What I'm doing is reflecting the existing command line into maven.
>
> On Mon, May 26, 2008 at 3:54 PM, Glen Mazza <[EMAIL PROTECTED]>
>
wrote:
>>
>>> I believe we have something like this already on the command-line
>> (-validate option[1]). Perhaps it would be better to build on this,
>>
> so
>>
>>> command-line/Maven/Ant users can use it as well.
>>
>> Glen
>>
>> [1] http://cwiki.apache.org/CXF20DOC/wsdl-to-java.html
>>
>> 2008-05-26 Benson Margulies wrote:
>>
>>> Watchers of checkins can see that I'm inventing a WSDL validation
>>>
>> maven plugin.
>>
>>>
>>> First catch:
>>>
>>> INFO: Resolve schema http://schemas.xmlsoap.org/wsdl/2003-02-11.xsd
>>> from baseURI:
>>>
>>
>> jar:file:/home/benson/.m2/repository/org/apache/cxf/cxf-common-schemas/2.1.1-SNAPSHOT/cxf-common-schemas-2.1.1-SNAPSHOT.jar!/schemas/wsdl/http.xsd,
>>
>>> namespace: http://schemas.xmlsoap.org/wsdl/
>>> [INFO]
>>>
>>
>>
>>
>>> [ERROR] BUILD ERROR
>>> [INFO]
>>>
>>
>>
>>
>>> [INFO] Non unique body parts, operation [ greetMe ] and operation [
>>> testTriPart ] have the same body block
>>> {http://apache.org/hello_world_xml_http/bare/types}requestType
>>>
>>
>>
>>
>
>>>
>>>
> ---
> Daniel Kulp
> [EMAIL PROTECTED]
> http://www.dankulp.com/blog
>
>
>
>
>
