Okay so great minds think alike i think, but looking at these two/three things:

1. touch $AS/standalone/deployments/example.war.skipdeploy
2. cp -r target/example.war/foo.html $AS/standalone/deployments/example.war
You are copying to a different directory, that is hard to GROK.

1. touch $AS/standalone/deployments/example.war.skipdeploy
2. cp -r target/example.war/ $AS/standalone/deployments
3. rm $AS/standalone/deployments/example.war.skipdeploy

It is a little odd to copy the new things skipped NOT into the example.war.skipdeploy directory?


1. rm $AS/standalone/deployments/example.war.deployed
2. wait for $AS/standalone/deployments/example.war.undeployed file to appear
3. cp -r target/example.war/ $AS/standalone/deployments
4. touch $AS/standalone/deployments/example.war.dodeploy
Above your touching a file you did not explicitly create above, again kinda odd.

I get what you are doing I think, and I like the directory renaming as I came up with it independently, but the seemingly magic/gaps from one step the next is troublesome in getting what I am able/should do?


Jim Tyrrell
Senior JBoss Solutions Architect

Did you see RHT on CNBC's Mad Money?
http://www.cnbc.com/id/39401056



On Apr 12, 2011, at 10:02 AM, Jason T. Greene wrote:

On 4/12/11 10:56 AM, Jim Tyrrell wrote:
Guys the way that we have worked in the past where exploded files just
works is a really good feature. One that definitely is a selling point.

Reading through the thread higher in the list in the forums I get why we
are here today.

However, burying this change in the META-INF directory seems like a real
pain from a usability perspective.

Take a look at this:

https://github.com/jbossas/jboss-as/raw/6a72ca965f01021d034049598451e0aca3ba38f3/build/src/main/resources/standalone/deployments/README.txt


What about renaming the exploded directory tree instead.

Copy
application.war
to the hot-deploy directory

rename application.war.dodeploy when you want to deploy it
rename to aplication.war.doundeloy when you want to undeploy it

WHen it is is deployed
application.war.deployed
application.war.failed with a outtext file in the directory that states
the reasons for failure?

If I do an lS in the directory I can immediately see what is going on. I
do not have to bury myself in META-INF directories and I would think
this is a lot simpler.

I did not read all of the pages, this seems like an elegant solution to
some of the "Atomicy" issues you have around deployment, along with
making this easy to use.

Does this mean that if I update an .jsp page in an exploded directory I
need to mark it as .dodeploy?

Have you guys talked with Max and the tools team to see what/how this
effects what they have to do?

Yes, I think hes happy with the eventual outcome. Although I will let him comment :)


Jim Tyrrell
Senior JBoss Solutions Architect

Did you see RHT on CNBC's Mad Money?
http://www.cnbc.com/id/39401056



On Apr 12, 2011, at 9:42 AM, Jason T. Greene wrote:

Yeah basically we had not finished the feature until Beta2 and onward. I
highly recommend grabbing the latest upstream.

On 4/12/11 10:18 AM, Howard Gao wrote:
Thanks Jason. I think I would use auto-deploy for most of the time.
If by default it is auto-deploy for non-exploded deployment, then I
think there is an issue with it. I deployed a simple MDB (packed in a
jar file, I think that's non-exploded) and I still have to add a
.dodeploy to trigger the processing of my MDB.

Or perhaps it has been just so changed? My as7 is built from the code
days ago.

Howard


On 04/12/2011 11:10 PM, Jason T. Greene wrote:
Actually allow me to clarify:

BY DEFAULT you do not need markers for non-exploded deployments.

You do need them for exploded. You can enable auto-deploy for
directories, and ignore using any marker, but you have the same casino
odds that you had in previous AS releases.


On 4/12/11 10:04 AM, Jason T. Greene wrote:
You dont need markers for non-exploded deployments (see auto-deploy in
deployments/README.txt)

On 4/12/11 9:51 AM, Howard Gao wrote:
Hi,

In AS7 it uses a bunch of 'Marker files' to control and indicate a
deployment. Comparing to previous AS 4 I found this is not so
convenient. For example if I want to deploy an EJB jar I need to
drop it
to the /deployments dir and create a .dodeploy marker file. And if my
EJB failed to deploy the marker file changed to .fail mark file. In a
debug process this deployment can be repeated many times, each time a
.dodeploy has to be manually added. And that's for a single EJB. If I
have 5 EJB jars deployed I need file marker files manually created.

In AS 4 I just drop it and done. I wonder what's the good reason for
those marker files? From a user's point of view, for what other
purposes
could it be to drop something to the deployments directory than for
deploying it right away?

Thanks
Howard

_______________________________________________
jboss-as7-dev mailing list
jboss-as7-dev@lists.jboss.org <mailto:jboss-as7-dev@lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/jboss-as7-dev







--
Jason T. Greene
JBoss, a division of Red Hat
_______________________________________________
jboss-as7-dev mailing list
jboss-as7-dev@lists.jboss.org <mailto:jboss-as7-dev@lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/jboss-as7-dev



--
Jason T. Greene
JBoss, a division of Red Hat