Fork me on GitHub
#pedestal
<
2016-05-17
>
donaldball02:05:59

(Oh my. FWIW, while I stand by the above, it was intended as a private gripe to another party. I am embarrassed that I accidentally published it here. Had I intended a public critique, I would have been much more thoughtful and respectful.)

tessellator13:05:55

@donaldball: I get the same error when using Tomcat with pedestal+component. Even using a server start/stop/start sequence with the default pedestal-service project has issues when using Tomcat. I would agree this is likely a bug. Would you like to open an issue or drop a note to the mailing list?

donaldball14:05:30

Thanks for the confirmation that it’s not something specific to my environment; I’ll do so.

christianromney16:05:29

@donaldball: the dissoc in the snippet above will cause a regular map to be returned instead of a record. you might prefer assoc w/ nil

christianromney16:05:13

also, it's good practice to make the calls idempotent so that calling start on a started component just returns this. ditto for stop on a stopped component

donaldball16:05:16

I go back and forth on the idempotence of lifecycle fns tbqh

donaldball16:05:38

But the dissoc in this case is preserving the type. Is this behavior new in clojure 1.8?

christianromney17:05:53

not new functionality (record dissoc). the reason i usually go for idempotent lifecycle is to avoid the exception that's thrown on start when you're already bound to the port and i never have any good reason for not making them idempotent