libera/#clasp - IRC Chatlog

11:58:43 drmeister bike: I've forgotten how the printer works.

11:59:23 drmeister Removing __repr__ causes the output to change to #<geom::m4 >

11:59:31 drmeister I think I need a `fields` method

12:05:19 yitzi __repr__ is called by write-ugly (or something) as the fallback to print-object. So either you have a print-object specialization or a __repr__ ... or you get print unreadable.

12:06:33 drmeister I removed the __repr__ function and I'm implementing a `fields` method that will generate and consume a vector of 16 numbers.

12:07:05 drmeister That should correct things.

12:19:16 drmeister Almost. With (eq *print-readably* t) I now get...

12:19:32 drmeister https://www.irccloud.com/pastebin/AhYw5nDY/

12:20:36 bike that actually looks pretty good

12:20:52 bike encode/decode are already defined, and they're used by #i

12:21:17 drmeister And with *print-readably* set to nil I get #<geom:m4 >

12:21:29 drmeister I'll add back the __repr__

12:24:28 drmeister This works...

12:24:30 drmeister https://www.irccloud.com/pastebin/Io702cJK/

12:26:06 drmeister Hmm, but it is bypassing the fields function I just added and using encode/decode

12:27:13 bike but encode and decode should already be correct.

12:27:32 bike so it's fine, right?

12:33:05 drmeister It works fine now - I just don't understand the flow of control. It doesn't appear to invoke the fields function. I'm digging a bit deeper to remind myself how this all works when it's in C++.

12:40:41 drmeister The General_O::encode invokes fields. Since I implemented OMatrix_O::encode it bypasses it.

12:44:16 bike this is another thing we should really clean up.

12:56:13 drmeister With the compact vector writer I get...

12:56:14 drmeister #i(GEOM:M4 (:M . #171DdfdxSe4huf0c$KRQI5Wr$So81M7eeoPmb$@@@@@@bFrRcXDjzxbLUEUVgOh0JXgSJ$PHucMBoI5c7@@@@@vgJSpi3j8dGE9nL$Z4@6pdGEUI$ncXCaFZ6HUV@@@lc0PcY@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ ))

12:57:10 drmeister Now geom:m4 objects survive save-cando/load-cando

14:59:38 chkhd ** NICK chkhd`

17:05:05 drmeister I'm getting an intermittent crash when building cando

17:05:06 drmeister https://www.irccloud.com/pastebin/RQaqNvet/

17:05:41 drmeister I have assertion tests turned on.

17:08:26 bike uhh. maybe it's the auto compilation? do you have that on?

17:15:14 drmeister Can you remind me, where do we set that?

17:17:29 bike there's an environment variable... but you should be able to see it in the running image by seeing if clasp-cleavir::*autocompilation-thread* is nil

17:18:19 drmeister It is running, there is a thread in that variable.

17:19:40 bike okay. so maybe the simple fun is getting swapped out for a compiled version, but bytecode_call is still happening for some reason.

17:23:53 bike and the quick fix would be turning off autocompilation

17:29:15 drmeister Ok.