freenode/lisp - IRC Chatlog
Search
13:10:56
malice`
Hi! I thought I could use #'asdf:test-system to test my system, but it turns out that it returns T even if some of tests fail. I am using fiveam. Is this the proper behaviour of #'asdf:test-system?
13:17:53
malice`
phoe: I guess you may want to look at https://github.com/MatthewRock/cl-trie/blob/master/cl-trie.asd
13:34:27
phoe
if one of the lines of a defmacro docstring begins with a #\( then the indentation and compilation breaks
13:41:50
ogamita
phoe: a design flaw: font-locking only goes backward so far. It doesn't try to parse the file from the start.
13:53:02
_death
it's not a slime issue, but an old emacs issue.. there was recently a lot of discussion about it on the emacs development mailing list
16:10:44
loke_beirc
I'm going to have a look at how the input field of this application works. But that's for tomorrow.
16:44:32
thodg
I have a hard time finding out which implementations are supported by PORTABLE-THREADS
16:49:50
shrdlu68
Looking at the documentation, it seems there no way to POST binary data using trivial-http?
17:00:57
shrdlu68
https://github.com/svenvc/s-http-client: "HTTPS in only implemented on LispWorks (where it was really easy)"--how does Lispworks do ssl?
17:07:36
shrdlu68
XachX: I've been working in an ssl in lisp, cl-tls. Would it be uselful for quicklisp?
17:07:37
_death
could also use ssl on implementations that provide it.. and then people may work to provide it
17:12:25
shrdlu68
Most systems ship with certificate bundles. It's a metter of configuring a client to find them. Otherwise ship with the bundle.
17:14:20
shrdlu68
_death: Different organizations have different demands in the vetting process, so yes of course.
17:14:52
shrdlu68
But if you pick, say, the bundle that the Mozilla foundation uses, that's a safe choice.
17:15:36
_death
shrdlu68: it's not about organizations, it's about the direct users of your program.. they are the ones to decide what they trust.. the fact that most don't even know about it doesn't mean you should "ship a bundle" in your security-related program
17:16:58
shrdlu68
_death: cl-tls is a completely low-level tls library, so it leaves the choice of ca-certificates to the user.
17:18:27
shrdlu68
But people don't typically go around changing which ca certificates their curl or browser uses.
17:20:47
shrdlu68
_death: Even so, the sysadmin will likely be installing a ca certificate and a client certificate rather than removing those that the browser ships with.
17:23:14
_death
shrdlu68: and the browser usually doesn't ship with any afaik, but uses the operating system's list
17:24:38
_death
shrdlu68: looks like mozilla does has a default bundle.. which I guess is disabled by those organizations
17:25:26
shrdlu68
Anyway, cl-tls does validation, all the user has to do is point out where the ca certificates are.
17:26:26
shrdlu68
I'm curious about how other implementations handle wildcard host names, especially those that leave that to the user.
17:27:01
_death
even as a single home user, it may a be a good idea to review all those sources (there are way too many of them, that's another problem).. at least important servers nowadays adopt PKP
17:27:49
shrdlu68
An implementation that accepts *.com + a rogue sysadmin installing "company" CA certificates = trouble.
17:49:45
minion
ecraven: guild: Lisp Guild, a place for exchanging tasks to be done: https://github.com/Lisp-Guild/lisp-todo/projects
18:15:23
emaczen
For some reason, parenscript isn't compiling. I am just trying to remove it and re-install
18:16:57
Bike
the quicklisp dist is the collection of systems from quicklisp, so including parenscript and alexandria and everything else
18:18:16
Bike
of course if you've completely removed parenscript i think you have to quickload it again.
18:19:12
emaczen
I just did (ql:uninstall :parenscript) followed by (ql:quickload :parenscript) and I'm still getting a compile error
18:22:20
aeth
I've gotten problems twice before with cl-sdl2 that could only be fixed by going into ~/.cache/common-lisp/ and removing the compiled files for it
18:23:01
aeth
~/.cache/common-lisp/whichever-lisp-version/path/to/files/foo.fasl (or whatever your implementation calls the compiled files, if not FASL)
18:23:52
aeth
If that's the problem, reinstalling it on Quicklisp won't fix it because it will be installed to the same path and so use the same cached files.
18:25:30
nyef
Tied for longest such directory name is "sbcl-1.3.14.141.dx-functionals-383079.2-9ddefed-dirty-linux-x64".
18:27:28
aeth
That's nothing... once I accidentally messaged Nick (instead of NickServ) my password. That retired that password.
18:28:06
cheryllium
Oof. I was gonna say, one of my fears is accidentally voiding my password in that way.
18:29:10
aeth
Give it enough years and every possible mistake will happen in things like sh or IRC commands that like to punish people for typos.
18:29:24
aeth
Fortunately, on Freenode, you can send your password on connect, so you don't have to message anyone to identify.
18:30:17
aeth
once I overwrote an IRC log file because I didn't quote the > when searching for a name via grep
18:30:26
cheryllium
you know how common it is for things to have a folder named bin? he was trying to delete one of those, but instead managed to nuke THE bin
18:31:24
aeth
That reminds me of that article where someone did "rm -rf /" on some server and someone else had to recover it.
18:31:40
cheryllium
one time we did a find+replace on a directory but we didn't exclude image files, so all the image files in that dir got corrupted
18:32:44
cheryllium
wasn't the digitalocean incident something like this? they deleted their production db...
18:35:01
cheryllium
anyway, to stay on topic: is there any difference between :foobar and 'foobar in lisp? I understand the latter to be a quoted symbol, I'm unsure of the former though
18:36:17
cheryllium
I notice they seem to be interchangeable when I'm loading things like with quickload
18:36:24
aeth
cheryllium: generally, it's a good idea to use keywords instead of symbols where e.g. Scheme might always use a symbol... because of how packages work
18:36:47
Bike
they have the same symbol-name, and that's what's used as the string. the package is discarded.
18:39:05
aeth
cheryllium: off-topic, but your ls not found story reminds me of this: http://www.lug.wsu.edu/node/414
18:43:06
aeth
(originally posted in 1986, copied from a place where it was last modified in 1996, uploaded to that site in 2006, and moved to a new system on that site sometime around 2011 when the comments begin)
18:46:35
aeth
cheryllium: Anyway, either you mess with packages (symbols) or you use keywords. In the REPL (like when using Quicklisp) it makes no real difference to use "" or ' or : or even '#:
18:47:48
aeth
(the last one is non-interned, like what you get from gensym, and the first one, the string-literal quotes, will probably be in all-caps if you want to use those)
18:47:50
cheryllium
I wonder if it could be simplified further, CL packages seem to be a common source of confusion
18:49:27
jackdaniel
cheryllium: this is a nice (and short!) read which can possibly reduce some confusion
18:50:04
aeth
The confusion with packages is when you have something like * that is probably going to be imported from CL into all packages. So you can think that keywords and symbols are basically equivalent, up until the point where you use a symbol that isn't in CL.
18:51:41
aeth
jackdaniel: (* 2 3) will work as expected even though it should probably be (:* 2 3) and you won't really notice until you also have a (foo 2 3) and suddenly your CASE or whatever you're using to process the symbols will only work internally for foo, because it is looking for your-little-language::foo
18:52:25
jackdaniel
if you "use" CL package, then symbols are imported from it, if you don't, then they are not
18:52:52
jackdaniel
so meaning of (* 2 3) depends on your current package (and its inherited symbols)
18:53:52
aeth
Right, what I mean is, the main point where you'd probably have confusion over symbols is when you're writing a fancy macro.
18:54:22
nyef
On the other hand, if you use DEFPACKAGE and forget the :USE term entirely, you may-or-may-not end up with a package that uses :CL.
18:55:10
shrdlu68
This got me a few times when I was starting out with lisp. Very confusing and alarming errors.
18:55:55
jackdaniel
for me CL package system works fine. I had some confusion at start with forward references to not defined yet packages
18:56:50
aeth
What I like about the CL package system is that you can encapsulate things via macros by interning a symbol to (or hardcoding the path to) an internal package.
19:13:24
emaczen
Where can I find newer documentation for parenscript? It seems to have changed a lot since I last used it.
19:23:19
phoe
unrelated - a recent screenshot of LambdaDelta 0.98.1 running actual LMI stuff http://i.imgur.com/D0sJJKR.jpg
19:49:45
nosefouratyou
when I run a function that has something undefined, why don't I get an error? in this example https://gist.github.com/nosefouratyou/7653375e28e74d106f68663099124e38 I didn't define message-parse but I don't get an error
19:51:13
phoe
so the undefined-function error gets caught by the handler-case and you do not get cast into the debugger
19:53:19
nosefouratyou
for some reason I thought that an undefined error wouldn't be caught by the exception handling mechanism. guess I underestimated how complete/awesome it is.
19:55:06
phoe
nosefouratyou: well, an error is an error in Lisp, which is a condition of type undefined-function which is a subtype of error
20:11:06
phoe
but seemingly "So far we have been able to run full-speed on a 2.3 GHz i7 and a 2.5 GHz i5, but the i5 was pushing it. The i7 gets a bit warm."
20:49:12
phoe
...I am really going to check how well the Windows version of SBCL behaves on Linux/Wine.
20:52:27
drmeister
(clos:subclasses* xxx) returns a list of all classes that inherit from class (including class)
20:53:29
phoe
So you don't need to cons up new lists - you can destructively append the ones you cons in the depths.
20:54:46
mhd
Hi, using SBCL. Trying to use defsystem, but forgot how to get defsystem (ASDF). How do I get defsystem in fresh SBCL? I did ./run-sbcl.sh and then (require :asdf). No workee. I know I could probably install quicklisp, but is there some way without using quicklisp?
20:57:43
phoe
If, in some weird case, you don't have it, https://github.com/fare/asdf has an installation guide.
21:04:30
mhd
This worked: download https://common-lisp.net/project/asdf/archives/asdf.lisp, and then LOAD it.
21:06:21
Xach
mhd: if you cannot (require :asdf), something is incomplete or broken with the installation.
21:08:39
mhd
I think it's just: be on Mac, download latest SBCL, run it from shell using ./run-sbcl, do (require :asdf) => error: debugger invoked on a SB-INT:SIMPLE-FILE-ERROR:
21:19:09
mazoe
mhd: FYI I have sbcl v1.3.16 installed and working fine on a Mac. Installed with Homebrew
21:36:24
mhd
_death: nothing I tried worked (improved things): tried ./ and tried absolute pathname to same and no workee
21:39:12
francogrex
hi what was the command in slime where the evaluation results gets printed inside (below) the .lisp file expression?
21:47:59
mhd
prxq: I'm trying to respond to _death's request to set SBCL_HOME appropriately and then try ./run-sbcl and see if (require :asdf) then works.
21:56:21
mhd
prxq: it's going from http://www.sbcl.org/platform-table.html => download latest for Mac => unarchive => cd sbcl-1.2.11-x86-64-darwin => INSTALL (text file to read for instructions) => follow these instr's: 1. BINARY DISTRIBUTION
22:24:13
mhd
prxq: xach: OK, solved: I had an old ~/.sbclrc init file around, which in my case changed the working directory using sb-posix:chdir. Running without init file works, among other possible things I could do once I realized this. I think it's pretty much not-a-bug.
0:17:49
pillton
kruhft: There are people in this thread that have used it: https://mailman.common-lisp.net/pipermail/pro/2015-August/001227.html.