freenode/#lisp - IRC Chatlog
Search
15:32:30
pfdietz
Once real time electricity pricing becomes the norm, there should be a Linux API for it, so I can scale the testing rate up or down.
16:33:12
pjb
While quickloading osicat, I get: Error while trying to load definition for system osicat from pathname /Users/pjb/quicklisp/dists/quicklisp/software/osicat-20180228-git/osicat.asd: There is no package named "UIOP/LAUNCH-PROGRAM" .
16:34:13
pjb
Note that: The most up-to-date client, version 2017-03-06, is already installed. 1 dist to check. You already have the latest version of "quicklisp": 2018-10-18.
17:36:53
jackdaniel
pjb: quicklisp won't pull uiop because it is already present on the system. you need to pull uiop yourself to somewhere where it will be picked from (i.e local projects)
17:37:29
jackdaniel
this is brain dead and on its own a very good reason why asdf shouldn't bundle uiop as exported interface (it could have its own copy asdf/uiop-internal or whatever)
17:40:26
jackdaniel
(in particular that means, that despite quicklisp shipping uiop compatible with cffi, asdf says "nah, I'm fine" and quicklisp doesn't download the dependency)
17:51:46
pjb
jackdaniel: that, and problems with .cache/common-lisp too. removing it and re-quickloading works.
18:19:01
pfdietz
If uiop is already present in the system, why doesn't it work? QL wants a different version?
20:46:49
pfdietz
The issue here, then, is that you want two different versions of UIOP to be available, but the package names collide.
20:56:16
aeth
pfdietz: No, afaik UIOP and ASDF are designed to be "upgradable" so that name collision issue is avoided. The issue, though, is that it doesn't detect when to upgrade, probably because only ASDF and UIOP do this.
20:58:12
aeth
pfdietz: In SBCL do (ql:quickload :uiop :verbose t) on a fresh image and you get a warning that you're redefining just about everything, and hence "upgrading" the old one to the new. Advanced macro magic. That's also why I now do (ql:quickload :uiop) (ql:quickload :foo :verbose t) instead of just quickloading foo. Hides so many warnings.
21:03:45
pfdietz
I wonder if there would be much demand for a facility that could compile/load files in a "package sandbox": if a package FOO were referenced, it would use something like <sandboxname>/FOO instead (with some way to escape out to the global namespace).
21:23:37
ebrasca
elderK: Fine, thinking how I can have 1 representation for reading 1 item and reading sequences.