Change cacheBust to boolean with default value true, and passthrough query parameter from phetmarks

[samreid]

cacheBuster saved us many times in the past while we were getting used to JS caching, but now it is somewhat of a hindrance because it is impossible to set breakpoints in the dev tools and have them persist across reloads. We can use a query parameter cacheBuster=false, but I'm wondering if this would be a good time to set cacheBuster=false by default. I'm wondering what will happen on iPad--perhaps cache buster is essential there.

[samreid]

I'm commandeering this issue based on today's dev meeting discussion.

@jessegreenberg said:

I have been tricked by browser caching a lot lately, especially on iOS Safari and Firefox. Would it be possible to enable ?cacheBust by default in RequireJS? Or possibly by default from phetmarks?

@jonathanolson replied:

This has been a pain point for me. I would at least prefer making it a default in phetmarks, as it is incredibly rare for me to want the caching.

I replied

It’s been very productive for me to have persistent breakpoints across reloads. I have to remember to run with dev tools open so I don’t get stale cache, but it has been worth the trouble for me. But maybe we need to make it so different developers can have different defaults? Perhaps a pass-through query parameter for phetmarks? Like if you bookmark http://localhost/phetmarks/?cacheBust then that query parameter passes through to launched sims?

@pixelzoom said:

+1 for pass-through query parameters, +1 cacheBust=true default

@jbphet said:

+1 for pass-through query parameters, +1 cacheBust=true default

@Denz1994 said:

+1 for cacheBust = true default

Consensus:

SR: I’ll make it boolean, default to true. There is a usage in phettest to change. I’ll pass through query parameters in phetmarks. Hopefully next week!
MK: Update logic in the dev HTMLs
CM: Update the config.js files, probably just doc?

[samreid]

The core changes are committed above, those commits will be most pertinent for review. Following are the commits for grunt update.

[samreid]

Note to self to consider #759 and phettest moving forward. Commits to many repos follow.

[samreid]

Marking for dev meeting public service announcement to indicate that cache busting is now back on by default, specified by a boolean query parameter. Pass-through query parameters for phetmarks not yet implemented, should be done in a separate issue, perhaps at different priority. Anybody else want to work on that part?

[jessegreenberg]

Thanks for working on this @samreid, I can work on the phetmarks part. Is it still beneficial to "pass through" to phetmarks or can we add an entry in the phetmarks query param list?

[samreid]

I'm not sure I would use that "pass through" too much, let's see if it is important for others.

[pixelzoom]

@samreid The way that this is presented in phetmarks is confusing, it's a checkbox whose value is !cachBust.

Can we please change to something like this?

• Cache Bust (cacheBust=true|false)

Another reason for changing is that the passthrough ?cacheBust value would also need to be inverted to set the current checkbox value.

[jessegreenberg]

Agreed its a little confusing. But when a checkbox above is unchecked it doesn't set a value to false it just allows default value for query param. Since default for cacheBust is true in initialize-globals we need the checkbox to disable.

[jessegreenberg]

So for meeting 11/14/19 there are two remaining questions.

Does anyone want the ability to pass a query parameter to phetmarks to populate values of the sim "Query Parameters" checkboxes?
Can we live with a "Disable Cache Bust" checkbox or should we work on changing this to a "Cache Bust" checkbox?

[samreid]

At dev meeting:

@zepumph recommends a new issue in phetmarks to properly support boolean query parameters in phetmarks, rather than just being able to append a string or not.

@pixelzoom also volunteered to clean up the checkbox label itself.

We also agreed not to implement the passthrough.