Improve (simplify) Primefaces (Ajax) / Bootstrap interaction

[scolapasta]

With Bootstrap, we do a lot of "binding" of html elements, with javascript, e.g. tooltips or popup previews for image files. This is done by calling a JS function bind_bsui_components when we ready the document.

With Primefaces, we do a lot of ajax calls that update parts of pages - when that happens the bootstrap binding is lost. This is because the update replaces the html, but doesn't know to call the rebinding.

We currently solve this by calling the javascript for each update with oncomplete="bind_bsui_components"

This is prone to error, because we have to remember to put it everywhere.

[scolapasta]

So, I opened this issue because I tried an experiment that worked! I think it needs some more testing / refinement, but something of it's nature should eliminate the need to include oncomplete="bind_bsui_components" for all updates.

In dataverse_template.xhtml:
add:

      <p:fragment>
        <p:autoUpdate/>
         ...
     </p:fragment>

around the <script> at the end of the page.

That said, @mheppler are a little unsure if why it works, as we weren't sure if the document.ready would or should get called when that part of the page us updated. So it would be good to understand better why this is working and make sure there's no harm in calling other aspects.

(in doing some other research, there's another thing i want to try and will report back on shortly.

[scolapasta]

OK, tried the other thing I had found:
https://api.jquery.com/ajaxComplete/

But it did not seem to work as I expected. More experimentation to be done, especially with the first solution, which seems to work.

[scolapasta]

I keep seeing things like:
"The $(document).ready() event fires once, when the DOM has fully loaded from the initial load. An ajax event may change the DOM, but it doesn't reload the page, so the $(document).ready() won't fire again."

In the experiment above, though, we are effectively redefining the document.ready by updating the part of the page that contains it, and that does trigger it to fire. But I have yet to find any agreement (or disagreement) that this is an ok approach.

That said, I may have found a more "accepted" way. Omnifaces has an o:onloadScript component that will get run after ajax calls:
https://stackoverflow.com/questions/14400624/jsf-primefaces-ajax-updates-breaks-jquery-event-listener-function-bindings
https://showcase.omnifaces.org/components/onloadScript

I tried it, and it seems to work as well; just added:

        <o:onloadScript>bind_bsui_components();</o:onloadScript>       

to the dataset_template.xhtml (and removed the bind call from the documnet.ready).

[scolapasta]

So to be clear, I think this issue is ready to be worked on:
• add the onLoadScript
• clean up the oncompletes

[mheppler]

Lots to clean up...

• dataset-license-terms.xhtml
• dataset-widgets.xhtml
• dataset.xhtml
• dataverse.xhtml
• dataverse_template.xhtml
• dataverseuser.xhtml
• editFilesFragment.xhtml
• editdatafiles.xhtml
• file-edit-button-fragment.xhtml
• file.xhtml
• filesFragment.xhtml
• guestbook.xhtml
• harvestclients.xhtml
• harvestsets.xhtml
• manage-groups.xhtml
• manage-guestbooks.xhtml
• manage-templates.xhtml
• metadataFragment.xhtml
• permissions-manage-files.xhtml
• permissions-manage.xhtml
• provenance-popups-fragment.xhtml
• themeAndWidgetsFragment.xhtml

Found some inventory documents (doc, spreadsheet) that might be able to help track the various features, across all the pages.

[mheppler]

Related code clean up Infrastructure Spring Cleaning - Remove misc files #5348.

[mheppler]

Added first official use of onloadScript to the dataset pg as part of Truncate methods: checksums, long summary metadata fields #6685.

<o:onloadScript>summaryDescTruncation();</o:onloadScript>

[djbrooke]

• The failure case is likely not breaking, but potentially ~millisecond slowdown
• Testing shouldn't involve all the files above - will include some guidelines as part of the PR
• We should make sure the things in this rebind are there intentionally - suspicion is that there will not be issues
• Verify that widget views still work (if there are things that do not use the template)