Enterprise Client Community

You can sign the jar on your own, that's what we do. Or you can wait for the next Update (and hope that the fix is included which I think it will).

Tobias

Hello,

we had a similiar issue when validating jar files. But in our case, the jar files were signed twice. Take a look at my post: http://www.captaincasademo.com/forum/posts/list/1831.page

Tobias

Hi Markus,

we are not using maven, but we use Ant and Apache Ivy for dependency management.
For managing Captain Casa dependencies, we add the latest Captain Casa versions dependencies to our company repository more or less manually. But other projects depending on Captain Casa can easily add the dependency.

Don't know if that helps. But I would love to see Captain Casa move into a direction where Maven and Ivy can be handled more easily.

Tobias

And by the way: you have to switch to Ant 1.8.3 or higher if you want to use the digestalg property.

Hello Community,

we have custom components integrated in on client side of Captain Casa, so we have our build server handle the signing of the jar files along with the other client side jar files that are part of the Captain Casa delivery.

After switching from Java 6 to Java 7 on our jenkins build server, suddenly our application did not start any more. The error message looked like this (the class does not matter):

java.io.IOException: invalid SHA1 signature file digest for javax/servlet/ServletRequestWrapper.class
at com.sun.deploy.cache.CacheEntry$10.run(Unknown Source)
at java.security.AccessController.doPrivileged(Native Method)
.....

We found out that the reason for this is that the default sigining algorithm was changed from SHA-1 to SHA-256 between Java 6 and Java 7. In some jars, the Manifest file contained two signatures, one for SHA1 and one for SHA256, like this:

.....
Name: javax/servlet/ServletRequestWrapper.class
SHA1-Digest: goVnVYplb8EGVWBCMZ5GWeBAOyM=
SHA-256-Digest: 3YDDRnyrJT9CDOOpt4ZGduda6FnsFuvT3UsOs+a1TUo=
.....

The solution was to adjust our jar signer to explicitly use SHA1 as digest algorithm. Just add "-digestalg SHA1" to your signjar call (in ant it would be the digestalg property of the signjar task).

Hope this helps you get around these issues.

Tobias

Just found this article about how to avoid it:
http://netbeanscolors.org/devblog/index.php?post/2013/02/09/Download-a-Java-Runtime-without-the-Ask.com-or-McAfee-ads

But I still signed the petition.

Tobias

Hi Björn,

I thought of the solution with the extra icon in the header too.
But the problem is that we have the drag&drop logic for removing search fields but we will have a "click-icon-in-header" solution for grid columns.

I will check if we do not use the drag&drop solution for removing fields at all.

Thanks,

Tobias

Hi Captain,

FixGrids have the general feature to support drag&drop of grid headers to change the sequence of the columns. This happens on client side and the information about the columnSequence is transferred to the server with the next roundtrip.

We want to make a search grid configurable to the user using an almost WYSIWYG administration page. The user can select a field from a list of fields and drop it onto the grid and the new column is rendered. No problem so far, even the column sequence can be extracted and persisted in our database.

For the search fields, which are also part of the search configuration we remove fields by dropping them from the WYSIWYG part onto the original list of fields.

Is there a way to have the same behaviour for GridCols (drag&drop that goes to the server)?

Tobias

Just as a reminder: Ivy is a good tool for managing dependencies. It handles different versions of the same lib correctly.

Remember my presentation on the Captain Casa meeting two years ago. ;-)

Tobias

Hi Jörg,

here a very simple form example:
<?xml version="1.0" encoding="UTF-8"?>
<form name="CardForm" layout="simple">
<description>Edit Card Data</description>
<panel id="1" name="First Panel" colspan="1" newline="false">
<group textid="Person" colspan="1" newline="true">
<field name="FirstName" colspan="1" newline="true" datapool="Card" editstatus="1" required="false" unique="false"/>
<field name="LastName" colspan="1" newline="true" datapool="Card" editstatus="1" required="false" unique="false"/>
<field name="BirthDate" colspan="1" newline="true" datapool="Card" editstatus="1" required="false" unique="false"/>
</group>
</panel>
</form>

We have some structural elements like panel or group and have some input controls like field.

We take this definition and convert it to a component tree on server side which will be put in the rowdynamiccontent.
The backing bean is an instance of some generic bean which knows how to construct the component tree out of our form definition and which knows how to handle the values map.

The backing bean is constructed in a subdispatcher of the root dispatcher to make sure we get a fresh instance every time a task is opened.

The reason we did not use the activiti form definition was that we wanted to be free to add our own feature (like searching the database by clicking a button and automatically filling the fields in the form using the result of the search operation).

Tobias

Hello Jörg,

do you really mean Activiti, the open source bpmn 2.0 process engine?

We use the Activiti engine in our product, but we use our own form definition and reference the name of the form in the process.

The form itself contains of fields which are also configured in our system. The form definition is dynamically rendered (using ROWDYNAMICCONTENT). The binding to the process variables is done using a map in the backing bean. The map is filled using the process variables when entering the user task. The user edits the data in the form (which changes the variables of the map in the backing bean). When the user clicks next, we take the variable map of the backing bean and put it in the process engine as process variables.

Tobias

Hello Community,

I suggest we should join the Sandkastenliga again for this years European Championship in Poland/Ukraine. Was fun the last times and since the winner was from vps twice, we want to continue this game. ;-)

What do you think?

Tobias

Thanks for your answer. And yes, we use ROWDYNAMICCONTENT, but the content nodes are build up of (sub-)types of ComponentNode. Right?

Tobias

Hi Captain,

do you have any recommendations about the maximum size of a jsp page? In general, we have small pages which often include other pages using rowpagebeaninclude. These pages load pretty fast in your layout editor (<10 s).
But now we created a large page containing around 170 elements which takes several minutes to load in the layout editor. During runtime, the pages loads fast, so no problems there, but it is really a pain to change the layout in the editor.

In general, is it better to compose a page from several other beans using the rowpagebeaninclude feature? (of course, modularity). But from a technical point of view? Or is it better to generate the component tree on server side (ComponentNode feature)?

Very basic question, but I just wanted to know if you have any recommendations.

Tobias

Hello Captain,

we have an autocomplete element performing a search to our database. Everything works fine. However there is a problem if the entries in the list have identical values (the text shown to the user), but different valueIds (currently we use the database Primary Key to be able to identify the entries).
Example: value id in brackets
Entry1 (123)
Entry2 (456)
Entry2 (789)

When I first select the entry with valueId 456 and in a second step the entry with valueId 789, the setter of the valueId is not triggered on server side. So even though I selected an other entry (with different valueId), the change is not transferred to server side.

Tobias