candlepin April 2010

candlepin@lists.stg.fedorahosted.org

9 participants
22 discussions

deploy script changes
by James Bowes 30 Apr '10

30 Apr '10

Hey all: I've added two new env vars that the deploy script can now recognize: FORCECERT - force all certificates to be regenerated HOSTNAME - override the CN in your certs Use it like: $> FORCECERT=1 HOSTNAME=localhost buildconf/scripts/deploy And no more CN=XX! -James

3 3

A Unit Testing + Guice Question
by Devan Goodwin 30 Apr '10

30 Apr '10

I would like to be able to have an individual unit test configure the Principal object that Guice will inject into the component I'm testing. Currently this is done in Guice Modules and thus doesn't really allow you to change bindings on a test by test basis that I can find. Anyone know of a way? I've got a workable solution but it's not real pretty, bind Principal.class to a TestPrincipalProvider (all exists today), TestPrincipalProvider calls out to a new singleton object to get the actual Principal. The individual tests can also get access to that singleton and configure the result they'd like to be returned. Singleton needs to be cleared in teardown. It seems like there *must* be a way to do this on the fly with Guice but I can't find it yet. I did some digging and found mention of some introspection capabilities in Guice but nothing I tried seems to work, and I've yet to locate any examples of anyone doing this. Thoughts? Devan -- Devan Goodwin <dgoodwin(a)rm-rf.ca> http://rm-rf.ca

3 2

On API Versioning
by Bryan Kearney 27 Apr '10

27 Apr '10

One of the backlog items talks about API versioning: (E2) As an application integrator, I would like to define what version of the API I support in order to be protected from API changes. The currenting thinking I am hearing is that you could use the media type for this, but it would be easier to just use the URL Path /candlepin/some/resource candlepin/v2/some/resource where no v implies v1. If folks are comfortable with that, it would mean no work until we do v2. I would move this off the backlog unless there were concerns. Comments? -- bk

7 9

RPM Building checkpoint
by Adam Young 27 Apr '10

27 Apr '10

I just had a successful build with all unit tests passing. In order to do this, I needed the version of buildr I created, and called buildr from the command line with: buildr -f buildfile.new -r localbuild.rb package localbuild.rb is checked in, but I haven't added my version of the buildfile: I'll attach it here. I've tried to keep the versions of the RPMs I use consistant with the ones from the original buildfile. The biggest diversion is quartz, which has a very heavy dependency list, and on versions of jakarta-commons that are not compatible with the rest of the project. This, I've used a much older version of quartz. If this proves to be a problem, I can battle through getting the more recent one built. These are the JPackage RPMS I have installed. These is a mix of 5 and 6 in there. jakarta-commons-logging-1.1.1-2.jpp6.noarch jboss-transaction-1.0.1-api-5.0.1-2.jpp6.noarch jakarta-commons-httpclient-3.1-1.jpp6.noarch geronimo-specs-poms-1.2-13.jpp5.noarch jboss-javaee-poms-5.0.1-2.jpp6.noarch geronimo-specs-poms-1.2-13.jpp6.noarch jdom-1.1.1-2.jpp6.noarch hibernate3-ejb-persistence-3.0-api-1.0.2-4.jpp6.noarch aopalliance-1.0-5.jpp6.noarch antlr-2.7.6-7.jpp6.noarch dom4j-1.6.1-12.jpp6.noarch asm-1.5.3-7.jpp5.noarch jakarta-commons-logging-jboss-1.1-4.jpp5.noarch hibernate3-entitymanager-3.4.0-2.7.jpp6.noarch jboss-common-core-2.2.14-2.jpp6.noarch glassfish-javamail-1.4.0-4.jpp6.noarch hibernate3-annotations-3.4.0-1.4.jpp6.noarch plexus-containers-component-api-1.0-0.a32.3.jpp5.noarch geronimo-ejb-3.0-api-1.2-13.jpp5.noarch geronimo-jta-1.1-api-1.2-13.jpp6.noarch jboss-common-logging-spi-2.1.0-2.jpp6.noarch glassfish-jaf-1.1.0-6.jpp6.noarch hibernate3-commons-annotations-3.1.0-3.jpp6.noarch jakarta-commons-modeler-2.0-5.jpp5.noarch cglib-2.2-4.jpp6.noarch I think the glassfish ones are required for building quartz, and are not strictly speaking necessary. This is an incomplete list of the ones I've built by hand: guice-throwingproviders-2.0-4.young.noarch guice-assistedinject-2.0-4.young.noarch jakarta-commons-cli-1.2-4.young.noarch jta-1.1-3.young.noarch persistence-api-1.0-3.young.noarch hibernate-commons-annotations-3.3.0.ga-1.young.noarch jakarta-commons-dbcp-1.4-3.young.noarch servlet-api-2.5-3.young.noarch jakarta-commons-collections-testframework-3.2.1-5.young.x86_64 guice-multibindings-2.0-2.young.noarch commons-dbcp-1.4-3.young.noarch hibernate3-validator-3.1.0-4.young.noarch resteasy-guice-1.2.1.GA-4.young.noarch hibernate-tools-3.2.4.GA-3.young.noarch guice-2.0-5.young.noarch jakarta-commons-collections-3.2.1-5.young.x86_64 guice-servlet-2.0-2.young.noarch emma-2.0-0.5312.5.fc12.noarch I only recently started tagging my builds with young, so I have others that were built with fc12. Here is a list of RPMS I've forced in with --nodeps. freemarker-2.3.15-1.jpp5.noarch.rpm jdom-1.1.1-2.jpp6.noarch.rpm dom4j-1.6.1-12.jpp6.noarch.rpm jakarta-commons-modeler-2.0-5.jpp5.noarch.rpm quartz-1.5.2-7.fc12.noarch.rpm geronimo-ejb-3.0-api-1.2-13.jpp5.noarch.rpm On 04/27/2010 02:49 PM, Jesus Rodriguez wrote: > Adam, > > I'm in the process of writing a script that will convert > your MySpecs tree into something I can use tito with to submit > the pkgs into koji.rhndev. > > Basically the structure would be one dir per project, for example, > emma would be turned into: > > MySpecs/emma > MySpecs/emma/emma.spec > MySpecs/emma/emma-2.0.5312-build_xml.patch > MySpecs/emma/emma-2.0.5312-code-source.patch > MySpecs/emma/emma-2.0.5312-dependencies_xml.patch > MySpecs/emma/emma-2.0.5312-no-javac-target.patch > MySpecs/emma/emma-2.0.5312-no-version-stamp-tool.patch > MySpecs/emma/emma-2.0.5312.pom > MySpecs/emma/emma-2.0.5312-src.zip > MySpecs/emma/emma_ant-2.0.5312.pom > > In the process of writing my script I found that some of the > Sources / Patches referenced by the spec files are NOT in > your SOURCES directory. > > For example, > > ls: cannot access SOURCES/buildr-buildpath.patch: No such file or directory > ls: cannot access SOURCES/heckle-local.patch: No such file or directory > > Do these need to be checked in? where are they? > > The reason behind what I'm doing is very similar to the thirdparty > tree we built for spacewalk. It allowed us to manage third party > packages which were not in fedora/rhel that we needed in koji.rhndev > to build spacewalk. This is the example same situation with the > packages in your MySpecs. This will all go away once we push > these upstream, but for now I think this will work. > > My plan was to: > > - get the script written (mostly done) > - fork MySpecs on github > - run the script > - prep repo for tito > - commit / push > - have you do a pull request to merge in the changes > > Thoughts? > > jesus > >

1 0

Re: On API Versioning
by Justin Harris 27 Apr '10

27 Apr '10

----- "Jesus Rodriguez" <jesusr(a)redhat.com> wrote: > On Tue, Apr 27, 2010 at 10:26:40AM -0300, Devan Goodwin wrote: > > The v2 in the URL is gross but it would help with keeping the code > > reusable I think as we could fork to different resources inheriting > > from common parent classes. Not sure if we could do the same with > > media type checking, at least not as easily. > > > > Would the plain URL always be the latest and greatest with vX for > > those who want stability? Or is it no vX assumed to be v1 and we > > increment for all future releases. > > > > based on bryan's email 'no v implies v1'. > > What will our deprecation policy be? mark it deprecated and > it gets removed in the next version? > > Also, having the v2 in front of the resources makes it a little > harder > to reuse code, but how many versions will we have? 2? 3? n? > I'm thinking no more than 3, personally 2 would be great. > > If the v2 was after the resource we could actually reuse the > same Resouce classes, i.e. > > GET /candlein/owner/v2/ > > This would simply be: > > @GET > @Path("/v2") > public Owner getOwnerNew(...) > > With the v2 in front of the Resource we'll have to create a new > set of resource classes. I would be very careful about this. Putting the version higher up conflicts with the REST semantics. I know the intent here, but convention would imply that that GET /candlepin/owners/v2 would be getting an owner named "v2". So that's a -1 from me. :) > > jesus > > -- > jesus m. rodriguez | jesusr(a)redhat.com > principal software engineer | irc: zeus > red hat systems management | 919.754.4413 (w) > rhce # 805008586930012 | 919.623.0080 (c) > +---------------------------------------------+ > | "Those who cannot remember the past | > | are condemned to repeat it." | > | -- George Santayana | > +---------------------------------------------+ > > _______________________________________________ > candlepin mailing list > candlepin(a)lists.fedorahosted.org > https://fedorahosted.org/mailman/listinfo/candlepin

2 1

Scrum status for Mon Apr 26
by Adam Young 26 Apr '10

26 Apr '10

Sorry for missing todays scrum,. Here is the Readers's Almanac^H^H^H^H^H^H^H^H^H^H^H^H^H^HMy scrum status for today. Still wokring on getting the build to complete using a JPackage repository. WHile the version numbers are now much closer than last week when I got it to compile, there is still a problem when I use a local m2 repo to store the MAven jars, regardless of where I fetch them. To reproduce: add the followng line to your buildfile and you should see the same thing: Buildr.repositories.local = "target/m2repository" This leads me to think that we have some sort of "bleed over" in the build process, whereby we are getting packages from a Jar not specified in the buildfile. Looking at the error message, it looks like the root cause is: ] Caused by: java.util.MissingResourceException: Can't find bundle for base name org.hibernate.validator.resources.DefaultValidatorMessages, locale en_US [junit] at java.util.ResourceBundle.throwMissingResourceException(ResourceBundle.java:1539) Which leads me to think that a JAR has been built without the .properities files for I18N. Today, I hunt this bug!

1 0

git commit messages
by jesus rodriguez 23 Apr '10

23 Apr '10

It would be helpful if folks could format their git commit message with a one line summary (80 char or less), followed by a blank line then a more lengthy detailed description (if necessary). For example, lot's of code cleanup, removed dead code. Removed unused imports Removed a bunch of unused code etc. This simple change will help make the automated changelogs easier to consume since I don't have to clean them up. Thanks jesus -- jesus m. rodriguez | jesusr(a)redhat.com principal software engineer | irc: zeus red hat systems management | 919.754.4413 (w) rhce # 805008586930012 | 919.623.0080 (c) +---------------------------------------------+ | "Those who cannot remember the past | | are condemned to repeat it." | | -- George Santayana | +---------------------------------------------+

2 3

Fwd: Re: [rest-discuss] Bussiness layer error handling
by Bryan Kearney 20 Apr '10

20 Apr '10

For when we talk about validation. -- bk -------- Original Message -------- Subject: Re: [rest-discuss] Bussiness layer error handling Date: Sun, 18 Apr 2010 10:57:26 +0200 From: Jan Algermissen <algermissen1971(a)mac.com> To: Jørn Wildt <jw(a)fjeldgruppen.dk> CC: Rest List <rest-discuss(a)yahoogroups.com> On Apr 17, 2010, at 9:22 PM, Jørn Wildt wrote: > Consider the situation where a client PUTs a new representation of some > business object, for instance an update to an existing customer. On the > server side the software detects that some mandatory fields are not > completed. How should this information be returned to the client? > > I want to be able to say more than just "400 Fields X,Y and Z needs to be > completed" - the error information should be encoded in a structured way. Beware that you cannot redefine the existing HTTP status codes. 400 means Bad Request and not something application specific. What you should do is choose the most appropriate HTTP error code and, as you suggest, include the details in the body. In your example, that would be 422 Unprocessable Entitye (== syntactically correct, but semantic errors) Jan > > My first shot is to define an error media-type like > application/vnd.error+xml in which I can embed all sorts of error > information. The the HTTP response could be: > > 400 Fields X,Y and Z needs to be completed > Content-Type: application/vnd.error+xml > > <errors> > <error>...</error> > </errors> > > > Is this on the right track? Are there any existing media types for this? > > Thanks, Jørn > > > > ------------------------------------ > > Yahoo! Groups Links > > > ----------------------------------- Jan Algermissen, Consultant NORD Software Consulting Mail: algermissen(a)acm.org <mailto:algermissen%40acm.org> Blog: http://www.nordsc.com/blog/ <http://www.nordsc.com/blog/> Work: http://www.nordsc.com/ <http://www.nordsc.com/> ----------------------------------- __._,_.___ Reply to sender <mailto:algermissen1971@mac.com?subject=Re: [rest-discuss] Bussiness layer error handling> | Reply to group <mailto:rest-discuss@yahoogroups.com?subject=Re: [rest-discuss] Bussiness layer error handling> | Reply via web post <http://groups.yahoo.com/group/rest-discuss/post;_ylc=X3oDMTJxMzU2ZjZrBF9TAz…> | Start a New Topic <http://groups.yahoo.com/group/rest-discuss/post;_ylc=X3oDMTJlcmZmMWtlBF9TAz…> Messages in this topic <http://groups.yahoo.com/group/rest-discuss/message/15298;_ylc=X3oDMTM2YTcxc…> (3) Recent Activity: * New Members <http://groups.yahoo.com/group/rest-discuss/members;_ylc=X3oDMTJmdjExYzE5BF9…> 1 Visit Your Group <http://groups.yahoo.com/group/rest-discuss;_ylc=X3oDMTJlb2pzdTM0BF9TAzk3MzU…> MARKETPLACE Stay on top of your group activity without leaving the page you're on - Get the Yahoo! Toolbar now. <http://us.ard.yahoo.com/SIG=15oq80ha6/M=493064.13983314.13965207.13298430/D…> ------------------------------------------------------------------------ Welcome to Mom Connection! Share stories, news and more with moms like you. <http://us.ard.yahoo.com/SIG=15om2cahc/M=493064.13814537.13965224.10835568/D…> ------------------------------------------------------------------------ Hobbies & Activities Zone: Find others who share your passions! Explore new interests. <http://us.ard.yahoo.com/SIG=15o8vk54s/M=493064.14012770.13963757.13298430/D…> Yahoo! Groups <http://groups.yahoo.com/;_ylc=X3oDMTJkY3NnY2kzBF9TAzk3MzU5NzE0BGdycElkAzQzM…> Switch to: Text-Only <mailto:rest-discuss-traditional@yahoogroups.com?subject=Change Delivery Format: Traditional>, Daily Digest <mailto:rest-discuss-digest@yahoogroups.com?subject=Email Delivery: Digest> • Unsubscribe <mailto:rest-discuss-unsubscribe@yahoogroups.com?subject=Unsubscribe> • Terms of Use <http://docs.yahoo.com/info/terms/> . __,_._,___

1 0

Owner / Subscription Certificate Upload Changes
by Devan Goodwin 19 Apr '10

19 Apr '10

Upload of certificate now requires authentication, presumably basic. I modified the deploy script to do this as guest:guest. If you have users defined in your candlepin.conf, this will break, you need to either add guest:guest here (auth.user.guest=guest), or remove all your users in candlepin.conf. (this turns DefaultUserAuthService into a passthrough, it will accept anything) If you do not have a candlepin.conf, or there are no users in it, no changes are required. When the cert is uploaded, it will be associated with the owner of the authenticated principal. I've modified the default user auth service to also create a "Default Org" if none exists, the first time someone tries to authenticate. This is also the assumed org for any users in candlepin.conf that don't have an org. (which you can add there by appending :org name to their entry) Let me know if you see any issues! Cheers, Devan -- Devan Goodwin <dgoodwin(a)rm-rf.ca> http://rm-rf.ca

2 1

hibernate-tools
by Adam Young 16 Apr '10

16 Apr '10

OK, I couldn't quite give up on hibernate tools, so I just spent one more day on it and...well, I got it to compile. I haven't tested it, nor any of the other RPMS for that matter. Hibernate tools was the only one I had to patch. Once I accepted that this was a viable route, I was happy. In the version of Hibernate that tools was origianlly built against, therei s a class called CacheFactory that basically pulls in a mother load of other dependencies. Really poor design. It also has four symbolic constants for strings. It was only these strings that the rest of the build needed, which means that this class wasn't even loaded at run time: javac replaces the strings with the literals anyway. So, I hacked everything but the symbolic constants out of the class and added in via a patch. diff -Nurd hibernate-tools-needed-classes/org/hibernate/cache/CacheFactory.java hibernate-tools-3.2.4.GA.new/org/hibernate/cache/CacheFactory.java --- hibernate-tools-needed-classes/org/hibernate/cache/CacheFactory.java 1969-12-31 19:00:00.000000000 -0500 +++ hibernate-tools-3.2.4.GA.new/org/hibernate/cache/CacheFactory.java 2010-04-15 19:04:03.091745593 -0400 @@ -0,0 +1,13 @@ +//$Id: CacheFactory.java 5685 2005-02-12 07:19:50Z steveebersole $ +package org.hibernate.cache; + +/** + * @author Gavin King + */ +public final class CacheFactory { + + public static final String READ_ONLY = "read-only"; + public static final String READ_WRITE = "read-write"; + public static final String NONSTRICT_READ_WRITE = "nonstrict-read-write"; + public static final String TRANSACTIONAL = "transactional"; +} With that, I should be done, but Dmitri snuck in yet another dependency on me today. I'll get that tomorrow.

5 6

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

candlepin April 2010