Ring enhancement and patches are best posted to the "ring-clojure" group. There's a chance they might be missed if only posted to the Clojure group.
Regarding wrap-reload, your enhancements would put it halfway between the current wrap-reload, and the wrap-reload-modified middleware available as a third-party library: https://github.com/weavejester/ring-reload-modified I'm therefore inclined to keep wrap-reload the way it is, and leave more sophisticated behaviour up to other middleware functions. Incidentally, are you familiar with the "lein-ring" plugin for Leiningen? https://github.com/weavejester/lein-ring The lein-ring plugin provides a command for starting a development web server. Any modifications to the source files are automatically reloaded using the wrap-reload-modified middleware. - James On 5 January 2011 16:21, David Jagoe <davidja...@gmail.com> wrote: > G'day, > > Ring ships with some development middleware that reloads the supplied > namespaces every request. This is fantastic for quick iterative > development, particularly because the JVM takes so long to start. > However as soon as you have more than a few namespaces every request > becomes rather slow - even if there are no source changes the > middleware reloads all supplied namespaces. I have tweaked the ring > reload middleware to only reload the namespaces if at least one file > below the supplied directory has been modified: > > > (defonce src-mtime (atom 0)) > > (defn read-src-mtime [dir] > (reduce > (fn [a b] > (+ a (.lastModified b))) 0 (file-seq dir))) > > ;; Copied and modified from ring.middleware.reload > (defn wrap-reload > "Wrap an app such that before a request is passed to the app, each > namespace identified by syms in reloadables is reloaded if any file > has changed below the identified source directory. Currently this > requires that the namespaces in question are being (re)loaded from > un-jarred source files, as apposed to source files in jars or > compiled classes." > [app dir reloadables] > (fn [req] > (let [new-mtime (read-src-mtime dir)] > (if (> new-mtime @src-mtime) > (do > (swap! src-mtime (constantly new-mtime)) > (doseq [ns-sym reloadables] > (require ns-sym :reload))))) > (app req))) > > There are two ways in which I can immediately see that this could be improved: > > 1. Don't pass in src directory as well as namespaces. If the user just > specifies the src directory then we can infer all of the namespaces? > 2. Don't reload all namespaces when a file is changed - only the > affected namespaces. I think this can become tricky. > > Anyway the current implementation works well for me and I thought it > might benefit others. Any feedback welcome. > > Can I provide a patch or pull request for ring? > > > Cheers, > David > > -- > You received this message because you are subscribed to the Google > Groups "Clojure" group. > To post to this group, send email to clojure@googlegroups.com > Note that posts from new members are moderated - please be patient with your > first post. > To unsubscribe from this group, send email to > clojure+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/clojure?hl=en -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en
