Also: how do you know if the cache is stale?<div> <div>Robby On Sunday, April 6, 2014, Spencer florence <<a href="mailto:spencerflorence@gmail.com">spencerflorence@gmail.com</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> How does this handle dynamic and lazy requires? <div class="gmail_quote">On Sun, Apr 6, 2014 at 11:30 AM, Greg Hendershott <<a href="javascript:_e(%7B%7D,'cvml','greghendershott@gmail.com');" target="_blank">greghendershott@gmail.com</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">For those following along at home, I just pushed to `experimental` branch an updated strategy: <a href="https://github.com/greghendershott/racket-mode/commit/0358c0248a062245c134c48053aed79298a71c3a" target="_blank">https://github.com/greghendershott/racket-mode/commit/0358c0248a062245c134c48053aed79298a71c3a</a> "When racket/gui/base not already loaded, examine the user module we're about to load to see if it will require racket/gui/base. If so, dynamic-require racket/gui/base on our main custodian. Then switch to user custodian and dynamic-require the module as usual. That way, we can do a custodian-shutdown-all on the user custodian, while leaving intact the main eventspace defined by racket/gui/base." How to "examine the user module we're about to load to see if it will require racket/gui/base"? Following the example of how syntax/moddeps works, recursively exploring using module-compiled-imports. (And memoizing, since there can easily be hundreds of imports to explore.) In my limited set of tests, this works as-desired. Of course there might be big holes in it I don't yet know about. ____________________ Racket Users list: <a href="http://lists.racket-lang.org/users" target="_blank">http://lists.racket-lang.org/users</a> </blockquote></div> </blockquote></div></div>