Posts Tagged With: Henry Newberry

Dirty pages and keeping users on them in #XPages

As I’ve mentioned before, sometimes XPages does not react the way your Notes users would expect it to react. In particular, I’ve had issues with ensuring that XPages warns the user when they try navigating away from a page that they’ve edited without first saving it. Such pages are “dirty”, since something has changed on them. Back in November we examined this in my post on Modified flags in XPages.

Unfortunately, sometimes I want to inform the user that they’re leaving a dirty page and other times I don’t. Also, sometimes, I seem to be able to add things that don’t warn the user that they’re leaving a dirty page. I’m sure there is consistency, but I haven’t figured out the rules, so I have added some code that handles these situations.

Simple warning of a dirty page

This was described in that prior post. Set the enableModifiedFlag to true and provide a couple of properties to handle it. This should fire whenever you close the XPage or navigate away via a standard link.

<xp:view xmlns:xp="http://www.ibm.com/xsp/core" enableModifiedFlag="true">
	<xp:this.modifiedControl>
		<![CDATA[#{javascript:if ( true ) {return "saveButton1"}}]]>
	</xp:this.modifiedControl>
	<xp:this.modifiedMessage>
		<![CDATA[#{javascript:"Purchase order modified. Click OK to save, cancel to continue without saving"}]]>
	</xp:this.modifiedMessage>
...
</xp:view>

Unfortunately, I’ve found that sometimes, it doesn’t warn me.

Warning when clicking links

I’ve got some navigation on the left side of my XPages, set up in a lovely dojo accordian with links to different parts of the application. However, not every link is set up in the same way. Some are simple and use the modifiedControl to warn users, like this one:

<xp:link text="Attachments" escape="true" id="link3" value="/att_attachmentManagement_view.xsp">
</xp:link>

Some, however, run some scripts to remove some sessionscope variables when switching pages and I was having challenges getting the server-side Javascript to execute. So, I found a way to do both:

<xp:link text="Payment requests" escape="true" id="paymentLink" target="_self" value="#">
	<xp:eventHandler event="onclick" submit="true" refreshMode="complete">
		<xp:this.action><![CDATA[#{javascript:clearDynamicViewSettings();
context.redirectToPage("/pro_paymentRequest_view.xsp");}]]></xp:this.action>
	</xp:eventHandler>
</xp:link>

The unfortunate thing is that when I’m using the redirect, it ignores the dirtiness of the page. Fortunately, Per Henrik Lausten found a way to deal with it on StackOverflow (thanks to PSolano), and posted it as an XSnippet on OpenNTF.

if (XSP._isDirty()){
  if (confirm ("Are you sure you want to navigate away from this page?" + "\n" + "\n" +
    "This document may contain unsaved changes." + "\n" + "\n" +
    "Press OK to continue, or Cancel to stay on the current page.")) {
    return true;
  } else {
    return false;
  }
} else {
  return true;
}

So, when I added that explicitly as client-side Javascript to each eventhandler, it worked beautifully. Now, being a fan of reusable code, I wanted to put it in a CSJS script library and invoke the function all over the place. My function was as shown above, with a little wrapper:

function isClean() {
	if (XSP._isDirty()){
	    ....
	}
}

Then, I just needed to invoke it. My first attempt looked great to me. Call isClean() and the client side would warn the user, then stop them as appropriate.

<xp:eventHandler event="onclick" submit="true" refreshMode="complete">
	<xp:this.action><![CDATA[#{javascript:clearDynamicViewSettings();
context.redirectToPage("/pro_paymentRequest_view.xsp");}]]></xp:this.action>
	<xp:this.script><![CDATA[isClean();]]></xp:this.script>
</xp:eventHandler>

Ummmm, but that didn’t work. In order to prevent the server-side Javascript from executing, the client-side Javascript needs to RETURN false, not just compute it.

<xp:eventHandler event="onclick" submit="true" refreshMode="complete">
 <xp:this.action><![CDATA[#{javascript:clearDynamicViewSettings();
 context.redirectToPage("/pro_paymentRequest_view.xsp");}]]></xp:this.action>
 <xp:this.script><![CDATA[return isClean();]]></xp:this.script>
</xp:eventHandler>

That’s added bonus knowledge – not only are we learning how to stick to dirty pages, we’ve also learned that if you have the CSJS return a value of false, it will not execute the SSJS. I’m sure you can grasp the opportunities for validation provided there.

What about when you don’t want to warn them?

In our application, the user can select the action to perform from a combobox and then click a button to execute it. (Thanks to Scott Good, Henry Newberry and the folks at Teamwork Solutions!) The challenge here is that when the user changes the value in the combobox, the page is dirty! I mean, no data may have changed other than that combobox, but the XPage still feels dirty. Since some of those actions will result in saving the document and all of them result in navigation away from this page in an expected and managed way, we don’t want it to feel dirty. To dodge the problem, I just add a little CSJS to convince that page that it’s not really dirty.

<xp:this.script><![CDATA[XSP._setDirty(false,"");]]></xp:this.script>

So, then the modifiedControl is not tipped off and we execute the intended commands. Now, I’m pretty sure we don’t run into this problem with standard save buttons, but we’re in an unusual case of taking two steps to execute one action.

Note

YMMV. These private calls to XSP methods may go away and using them requires care to ensure you’ve got the right parameters, or so the XPages Portable Command Guide advises us. Of course, everything we code may call functions that will go away and we always need to get the parameters right. So, I’m gonna use them, though sparingly.

Advertisement
Categories: Client-Side Javascript, Server-Side Javascript, Xpages | Tags: , , , , , , , , , , , , | Leave a comment

Puzzling behaviour of Java class files in design refreshes in #xpages

This is not a post in which I’m able to dispense some knowledge I’ve gained, but, rather, one in which I share my befuddlement by some of the new things I’m dealing with in XPages.

We have a database suite originally designed for us by Scott Good, Henry Newberry and the good folks at Teamwork Solutions. It’s based off their brilliant Process It! workflow Notes solution. In the original design, they used version 1.3 of Mark Leusink’s Debug Toolbar. Since version 3.01 is now available on OpenNTF, I decided to upgrade. One of the advantages of the upgrade is that now the code resides in Code/Java, accessible more directly from the Designer client than it was when it had to be stored in WebContent/WEB-INF/src. For me, this is far more convenient, since I can see more properties directly (especially about prohibiting design refresh).

Nonetheless, I seem to be having troubles.

I updated the design, then did a clean and a build on my template. Then, I refreshed design on the server copy from the template. At first, it seemed like it wouldn’t bring the built .class files over to the server copy from my template. Now, when I do a design refresh, it “horks” the five files, reducing them all to -1 bytes. If I do a clean & build on the server copy, the files get their expected size.

I believe my classpath is correct, as it builds them properly when I tell it to build in either the template or on the server.

<?xml version="1.0" encoding="UTF-8"?>
<classpath>
<classpathentry kind="src" path="Local"/>
<classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER"/>
<classpathentry kind="con" path="org.eclipse.pde.core.requiredPlugins"/>
<classpathentry kind="output" path="WebContent/WEB-INF/classes"/>
<classpathentry kind="src" path="Code/Java"/>
</classpath>

In the ACL, this ID has Manager rights, including for creating LotusScript and Java agents. I’ve signed all the files in Code/Java with this ID. I build it with this ID. I conduct the design refresh with the same ID and those same rights.

Does anyone have any ideas what I might be doing wrong that’s inhibiting the class files from refreshing with the rest of the design?

There are two Stack Overflow questions that deal with this directly:

Class files not being created after refresh from template?

XPages Java corruption issue

One of my co-workers did open the design to check the binding for one field, which must have corrupted the files once. That doesn’t explain all the instances, but it helps. This straddles the line between a reason to use templates (so no one even glances at the design in production) and reason not to use templates (because it’s forced me to do builds on production since template changes only produce corrupt class files). Interestingly, it’s only the Debug Toolbar class files that have the problem – the Teamwork Solution ones haven’t corrupted.

Update: It gets worse. I tried building on the local replica so that I could do it quickly, then replicate the class files up to the server. No luck. It overwrites the local files with the corrupted ones from the server.

Categories: IDE, Java, Utilities, Xpages | Tags: , , , , , , | 3 Comments

Blog at WordPress.com.

%d bloggers like this: