freenode/#lisp - IRC Chatlog
Search
10:57:54
schweers
Depending on the order in which I have my test cases, one branch or the other is claimed to be taken only one way. So some data seems to be lost?
11:00:29
schweers
I’m off for lunch for now, so sorry for not following up on any answers which may come ;)
11:04:33
scymtym
schweers: are you looking at the coverage of the test code itself? some test frameworks compile the test code when running the tests instead of ahead of time
11:56:28
flip214
> pkhuong> | I should also commit my fix to make SB-COVER obey in-package roughly the same way as swank
11:57:05
schweers
I have no idea. I just stumbled upon the coverage tool a few days ago and really like it.
12:04:19
loke
and there are about 300 billion different OBJ formats. :-) Which one are you thinking of?
12:05:06
flip214
loke: No, a 3D file format (from Stereolithography) -- nowadays used for 3D printing
12:12:05
flip214
where can I see what the source location of https://quickref.common-lisp.net/stl.html is?
15:30:33
akr
Hi, once I'm at a breakpoint in the debugger and I see the call stack, how do I examine the arguments with which a call in the stack was performed?
15:51:52
schweers
I wonder why people (myself included) find it harder to read all caps text than all lowercase text.
15:52:27
verisimilitude
I believe the explanation is we're more accustomed to lowercase and so better at recognizing it.
15:53:08
verisimilitude
That was my second guess. In any case, I can't link to the source, as I don't recall it.
15:53:27
sjl_
I like the all caps when looking at a repl session because it's easy to distinguish my input from the output. But it is pretty miserable to read large blocks of it like that.
15:54:27
verisimilitude
It makes for a nice delimiter for when I want to refer to Common Lisp. If I write in uppercase, it's rather easy to know which symbol I'm referring to.
15:56:40
verisimilitude
I recall being criticized for using it in my documentation, _death; I was told it belongs in the 1980s or something or another to that general effect.
16:01:06
_death
verisimilitude: don't take it too seriously.. opinions vary and there is an abundance of peanuts ;)
16:03:20
verisimilitude
Now since you're here, _death, I was wondering about that library you've been writing. What's the progress on that?
16:03:23
_death
verisimilitude: in general I wouldn't use &aux, because I don't think the lambda list is the right place for such variables.. but style flows and ebbs with years and programs..
16:07:36
_death
verisimilitude: well, I did touch it since our conversation, but mostly to pretty up some interfaces.. the pace for personal projects is the snail's..
16:09:01
_death
verisimilitude: sure.. in the meantime, it would be helpful to set up a git repository of your libraries though
16:10:36
verisimilitude
Well, I try to not release anything breaking. What did you have an issue with in particular?
16:11:16
verisimilitude
I've been trying to mostly add to ACUTE-TERMINAL-CONTROL, rather than change what's already there in a way that would break something.
16:11:55
_death
verisimilitude: I think you removed the mouse enable/disable function names from the export list
16:13:05
verisimilitude
Oh; I did do that, yes; I decided it wasn't reasonable to provide that. I generally try to avoid doing that, though.
16:16:28
_death
verisimilitude: I suggest to give it a try.. it's a complex tool but if you have the right model it's fairly simple to understand, and most of the time it's fairly simple to operate..
16:17:18
_death
verisimilitude: for a simple model, I think https://jwiegley.github.io/git-from-the-bottom-up/ was helpful
16:18:05
_death
verisimilitude: but for simple use, just create a directory, "git init", add some files, change them, and use "git-gui" to commit etc... and yes, magit is great too
16:19:51
verisimilitude
I've taken a quick look and everything currently in ACUTE-TERMINAL-CONTROL will stay, certainly. There's a PROPERTIES and (SETF PROPERTIES), but that's not documented because I'd like to have a better name and I'm not certain how I should best expose that, if at all.
16:20:28
verisimilitude
Considering these libraries are single files, I'd rather avoid this, _death, but I appreciate the advice in any case.
16:22:31
_death
verisimilitude: I'm guessing you associate version control with heavy set ups (servers, multiple authors) and this is why you wish to avoid it
16:23:21
verisimilitude
I'd rather avoid learning an explicit worse-is-better version of a decent file system.
16:25:30
verisimilitude
That's the gist of the present, I suppose. I know I should probably start doing something different.
16:26:00
verisimilitude
I've been told that my stuff is neat, but if I'm not on Github or some other big site it will mostly go ignored, and that's true, unfortunately.
16:26:06
_death
verisimilitude: just take an hour or so playing with chapter 1/2 of https://git-scm.com/book/en/v2
16:27:52
_death
verisimilitude: it's not necessary to use github, although it makes things simpler.. you can set up your own git serving daemon, or even a web front.. (you don't even need that; git can use http to clone a repository, but...) it's not a big deal
16:29:29
verisimilitude
Well, unlike everyone else who's told me to do this, you're actually using some of my software, so I'll consider it, _death.
16:30:30
_death
verisimilitude: sure.. maybe you'll find that it's very convenient (especially with magit...) to use git for any text that you happen to be editing
16:31:17
verisimilitude
As with most software, I'm going to want to use it as little as possible, _death.
16:31:24
schweers
git can even be useful locally, if you are the only person using it, and not sharing it across any machines.
16:34:47
verisimilitude
In any case, SHUT-IT-DOWN and CL-ECMA-48 are effectively finished and I could leave them as is, even though I'll occasionally make minor improvements, so I'd only be using git for ACUTE-TERMINAL-CONTROL if I decide to use git.
18:39:49
verisimilitude
I could certainly leave them, but it would greatly simplify and otherwise improve the code to do away with them.
18:42:44
_death
for my purposes I define something similar to a framebuffer, that only outputs the changes when REDISPLAY is called..
18:46:02
verisimilitude
I wanted ATC to complement Common Lisp, so I always preferred using something already present. Your library is mostly adding predefined interface components, right?
18:48:30
_death
currently what goes into it is not so well-defined.. but there are modules like that framebuffer abstraction, that can be used on their own
18:51:02
_death
for example, here's a small "demo" to test redisplay performance https://gist.github.com/death/c08917417b7acef288dcd28e9eb2c440
18:51:25
verisimilitude
There's a similar lack of constraint in ATC; I decided I'd mostly be putting in building blocks more than anything, so it's nice to know you're building such higher things in a different library.
18:58:54
_death
you can't do anything much with CL:TIME.. it just prints implementation-dependent text
19:00:19
_death
this program tells me how many redisplays happen per second (each benchmark takes 10s)
19:01:07
verisimilitude
I see you went with a WITH macro design, then; that's something I specifically wanted to avoid, not that I dislike them.
19:06:50
_death
it actually sets up the terminal for drawing of characters etc., and tears it down on exit
19:09:10
_death
still, with the addition of the input layer, which also uses acute-terminal-control, the two are the low level abstractions in the library..
19:10:02
_death
oh, the input layer doesn't directly use a-t-c either nowadays.. so it's the only module
19:10:54
_death
there is a drawing primitives module, and the rest is currently a salad of abstractions waiting for re-factoring
19:37:15
fouric
does anyone know of the top of their head how SBCL does (dis)assembly? specifically, does it use its own database of textual assembly <-> opcode mappings, or someone else's?
19:37:34
_death
verisimilitude: and I encourage you to continue with development of a-t-c :).. I think it's important to have a non-FFI alternative
19:38:50
_death
fouric: it has its own.. although there is now an SB-CAPSTONE contrib for some reason..
19:42:36
verisimilitude
Perhaps I'll take a look at Ncurses' API to see if there's any interesting functionality I could implement, since I'm not spilling over with ideas about it.
19:45:59
_death
verisimilitude: back then I inquired about urxvt.. I had to work around things a bit in the input module (and didn't bother with mouse events yet, as they seemed problematic)
19:50:45
verisimilitude
You've read the comments in ATC, though, I suppose, so you know my thoughts already.
19:52:06
verisimilitude
Since I'm using standard Common Lisp as much as possible, I'm unable to have special code for special terminals, so I can only target the LCD without extra facilities.
19:52:34
verisimilitude
Still, Xterm and its spawn did manage to agree on the sequences sent for the first twelve or so function keys and whatnot, I believe.
20:02:48
_death
I no longer remember.. is this info available in terminfo? if so, the database could be used
20:03:23
verisimilitude
You misunderstand; the issue is that you can't determine which terminal is being used in standard Common Lisp, as you can't access the environment in standard Common Lisp, _death.
20:04:47
verisimilitude
I still have a good amount of functionality I can add before I hit any such walls and must make a decision, I figure.
20:13:59
_death
maybe I should play with such a design in my library.. instead of my current urxvt-specific hacks
20:15:10
verisimilitude
Perhaps; another reason I wouldn't use terminfo or termcap, my sense of aesthetics aside, is because that's POSIX nonsense you can't expect everywhere.
20:18:04
verisimilitude
I've tried submitting patches for some awful issues to terminal maintainers or otherwise raising them, but with little success so far.