freenode/#lisp - IRC Chatlog
Search
9:52:38
phoe
I poke my nose into code that I haven't touched for months and I can quickly remember how it works and edit it.
11:17:27
Xach
scymtym: it takes 1.5 times longer than 1.4.4. i am looking at individual project timings now.
11:19:19
scymtym_
Xach: interesting, thanks. could we continue this conversation in #sbcl so we won't have to repeat everything for the others?
13:04:59
dxtr
Possible stupid question: Has there been any community efforts to create a new, "cleaned up", lisp specification? Seeing as a lot of stuff is marked as "deprecated" and things like that
13:05:59
dxtr
I don't specifically mean to create a new "web 2.0 web-scale all-the-buzzwords" thing -- but rather to clean up the specification and rationalize things
13:06:08
jackdaniel
cdr is meant for extending standard for missing features: https://common-lisp.net/project/cdr/
13:06:47
jackdaniel
cl21 is an effort to add some syntactic sugar and remove deprecated features: http://cl21.org/
13:07:24
jackdaniel
ergolib is another project which provides more intelligible package: https://github.com/rongarret/ergolib
13:09:05
jackdaniel
eulisp is a language designed to provide modern Lisp somewhere between Scheme and CL
13:09:21
scymtym_
beach has a project aiming at producing a corrected and more precise but compatible specification for common lisp
13:10:06
jackdaniel
there are many compatibility layer projects which bring various implementations together: bordeaux-threads, closer-mop, trivial-gray-streams, usocket
13:22:26
milanj
I don't think it's safe to set/get in hash-map without lock (even if you are using (current-thread) as key)
13:26:16
Shinmera
It's unfortunate that it's not possible to write a wrapper library to provide portable thread-safe hash tables (without shadowing CL).
13:29:10
shka
but since BT does not offer atomic operations, such hashtable may have bizmo performance unless it uses atomics from SBCL directly
13:31:47
jackdaniel
ecl's safe hash tables do the same (uses locks), but if I had a shared ht between threads I'd carefully lock code blocks instead
13:42:08
beach
dxtr: This is my project for improving the Common Lisp specification: https://github.com/robert-strandh/Well-Specified-Common-Lisp
13:51:06
beach
dxtr: It is not very ambitious, and I think the low ambition is required in order for it to have any chance of success.
13:59:02
shka
quote of the day -> "It is not very ambitious, and I think the low ambition is required in order for it to have any chance of success."
14:00:47
beach
shka: I maintain that. I think something like cl21 is way too ambitious. Same thing with efforts to include many more features.
14:04:28
beach
Yes, that's the idea. But already by specifying lots of unspecified behavior, it means many more programs can be portable across implementations.
14:05:22
beach
The next step (after this one) could be something like specifying the exact meaning of type declarations at different safety levels. But that is already too ambitious for the first step.
14:07:24
beach
But I maintain that the goal of WSCL is much higher priority than including more features or modifying the syntax.
14:09:24
jackdaniel
actually I believe Lisp could be better with some cruft removed (not necessarily added), but happily creating new package with limited set of symbols is not something what would require intervention on the implementaiton level (vide ergolib)
14:09:30
pjb
I would think it'd be better to use our own sexp-based structure. This would allow for a more formal approach to generate the standard.
14:10:14
pjb
It would also make it easier to write tools, such as to navigate cross references, (original paragraph), and to trace and reference things.
14:11:46
jackdaniel
well, most people know letters (probably all lispers!), let's use some syntax involving letters and they'll understand the semantics (promise)
14:11:48
pfdietz
The new feature I want to see become de facto standard in CL is package local nicknames.
14:12:35
jackdaniel
rme isn't against accepting pull request for it, so it is a matter of implementation
14:13:59
pfdietz
While poking through ccl, I noticed ccl has a very nice coverage facility included, with rollback. I hacked COVER to do that, but it's nice to have something integrated with the compiler.
14:24:15
jmercouris
I find it interesting, I hear people on IRC and I always try to imagine what they look like and what they speak like
14:24:37
jmercouris
I have these fuzzy images in my head for many people on this channel, as to how I imagine they are
14:24:51
beach
Type my name to the Google search engine and you will find out. It gives some Youtube videos of talks and such.
14:24:52
jmercouris
like when reading a book you come up with a face for a character, for me it is the same thing
14:27:22
dxtr
beach: To be fair if I google your name I get a whole lot of people with that name. 334 people to be precise :)
14:28:14
jackdaniel
ACTION also tries to figure out, what his accent/nationality has to do with lisp
14:28:36
jmercouris
It doesn't have to do with Lisp, but we are humans after all, and that is interesting
14:29:20
jackdaniel
jmercouris: of course, so asking such question once is fine, but having prolonged discussion about that is not. apparently he is not interested in such chit-chat, otherwise he would be there ;)
15:27:40
jmercouris
yeah, we are more than just a group of lisp robots, we are a community of people, and I don't think it would hurt to explore that territory a little
15:28:22
makomo
and always like to think about what that certain person was thinking during the period of time he was working on that project
15:29:54
jmercouris
I have no problem adhering to the rules, but I would like to understand the justification
15:30:12
makomo
well probably because there are a lot more people listening to the personal stuff here
15:30:29
beach
I think many people here would like to keep the discussion to the topic, and I am one of them.
15:30:58
beach
And that is why I keep repeating that my personal life is irrelevant and uninteresting.
15:31:02
jmercouris
Xach: The language is not *just* the definition, it is also its userbase and the applications and domain spaces they work in
15:31:37
Xach
I am happy to talk about Common Lisp applications and the challenges faced by people trying to use Common Lisp.
15:31:57
Xach
I do not want to talk about what time of day it is wherever you are on the planet, or what your weather is like, or whether you have exams coming up.
15:32:09
jmercouris
Then you would be also happy to talk about people, as the common lisp in their life does not exist in a singular bubble
15:32:36
beach
jmercouris: Stop it please. We already have an agreement about what is on topic here.
15:33:08
jmercouris
Ok, I concede :D, I still can't say I agree, but if everyone else does, so ist das
15:34:56
makomo
but i guess a *bit* of chit-chat here and there is fine though? i.e. just a couple of lines and not a full-blown discussion?
15:35:29
Shinmera
People will tell you when to stop. You just have to have the courtesy to actually stop when asked to.
15:35:55
dim
makomo: well trust people to tell you when you're off topic, and when you are, just say you're sorry and be silent for awhile… at least that's how I do it, apparently it kind of works ;-)
16:12:09
Xach
People still get paid in 2018 to work with and on Genera stuff, so it's still something.
16:13:15
Xach
It isn't of particular interest to me personally, but it's not especially dead compared to other dead Lisp things.
16:20:16
beach
Lycurgus: Have you seen this page? http://metamodular.com/Common-Lisp/suggested-projects.html
16:30:47
beach
makomo: I am guessing one reason is that they systematically choose programming languages that are not well suited for such programs.
16:31:20
makomo
each one misses something. also, it's pretty much a rule that they are rarely scriptable
16:32:08
makomo
beach: idk. for example, ghostscript has a feature for adding new bookmarks (although it's not really easy or friendly to use)
16:33:23
Xach
pdf is a pretty simple format in many respects, if you disregard the javascript stuff that was added. cl-pdf is interesting for parsing and generating pdfs. but it is quite low-level.
16:33:49
makomo
then there's also pdftk or w/e it's called, but that is pretty much deprecated since it uses GCC's Java implementation which has been removed from GCC 7 and up
16:34:23
makomo
that tool was quite versatile but there you go, not easily available anymore (at least for me on Fedora)
16:35:16
jmercouris
Xach: "pdf is a pretty simple format", the specification is 745 pages: https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/PDF32000_2008.pdf
16:35:19
schweers
makomo: really, they used the java frontend for GCC? Do they have plans on what to do, now that its been removed? It should run on the jvm, right?
16:36:28
makomo
schweers: no idea, here's the webpage: https://www.pdflabs.com/tools/pdftk-the-pdf-toolkit/
16:36:50
makomo
you'll want to go to "PDFtk Server" which is actually a command-line version of the tool
16:38:51
makomo
Xach: i was looking at cl-pdf and cl-typesetting. have you ever used/evaluated cl-pdf?
16:39:26
beach
makomo: I did not mean to imply that when a good programming language is chosen, the application is automatically fantastic. I just meant to say that when a programming language is chosen that is not suitable, it is much harder to create features that might be somewhat complex to implement.
16:39:29
schweers
does learning and using {Mc,}CLIM spoil one’s character in a way similar to learing and using lisp?
16:40:43
makomo
the interactivity that a lisp would provide would be a big bonus, but there isn't even a comprehensive library for editing pdfs in a non-lisp language
16:40:49
beach
schweers: It did for me. Though I never actually had the courage to write GUI applications using any other library.
16:41:24
beach
schweers: I started working on McCLIM because I couldn't see using an FFI-based GUI library for Gsharp.
16:41:55
schweers
I’ve never done much GUI. I did a little with borland C++ builder back in school and later java and swing once during my studies. Plus looking a little bit at gtk, but never really used it.
16:42:30
schweers
What I was getting at: is using CLIM that much different than using (say) gtk or Qt?
16:43:52
jasom
makomo: I use mupdf bindings for modifications, but for reading pdfs, I just add features to https://github.com/jasom/pdfparse when I need them
16:44:13
schweers
Nah, win32 seems to be really nasty. I’ve got a book on the subject, and still never really used it, because its really painful
16:48:24
jasom
beach: right, what primitives are needed for McCLIM; does it use native buttons/fields/&c. or does it draw them itself?
16:50:26
beach
jasom: CLIM was meant to work with native toolkits in addition to its own, but we never go around to doing that. It didn't make much sense.
16:53:02
jmercouris
jackdaniel: "professionally" as in a "profession" as in recieving money, I think you'll have a hard business case selling ncurses software in 2018
16:53:53
jmercouris
I wouldn't, but I'd definitely be surprised at how many were funded as new projects in 2018
16:54:00
jackdaniel
that said what I mean: I hack on things which are interesting to me and it happens that some of them let me sustain myself
16:54:01
jasom
jmercouris: there are other ways to make money with software besides selling it (we have in-house stuff with a curses UI for example, but the people who wrote it were clearly doing so professionally)
16:54:14
beach
jmercouris: I think jackdaniel, unlike most other computing professionals, actually spends a lot of time learning the tools and techniques that can profit his core business. Strange, huh?
16:54:38
schweers
I guess something curses like makes sense for some special applications, like for a cashier, or in a doctors office
16:54:44
jmercouris
no doubt he invests in himself and his tools, but that does not make a business case for an ncurses backend
16:55:52
Xach
makomo: I have used cl-pdf quite a bit for various things. i enjoy it and it is useful but it does require building higher-level abstractions yourself if you need them.
16:56:42
jackdaniel
jmercouris: either way I'm not doing it as a hobby - I'm working on CL full time and ncurses backend is made in that
16:56:54
jasom
Xach: cl-pdf loads the entire object stream into memory though, and in a rather inefficient manner. The pdf I wrote my parser for needed more than 4GB of heap to open in cl-pdf
16:56:54
jmercouris
I'm not trying to demean the project, but it is important to recognize that it appears very much still in the early stages, and is a hobby project
16:57:21
jmercouris
when it matures a little more, and you can start selling software you build with it, then you may say that you work on it professionally
16:57:22
Xach
jasom: yeah. there are definitely limitations. i was using it to automate pdf form filling.
16:57:41
Xach
jasom: the forms were usually a page or two so in-memory wasn't a problem (it was more of an advantage when poking around)
16:58:46
jackdaniel
as I said: working on something professionally doesn't mean necessarily selling it. I could mention reputation which directly impacts your chances, experience which directly impacts your chance of succesful delivery and other things. but it is offtopic I suppose
16:59:59
jmercouris
jackdaniel: There are tangential benefits for yourself as a professional, but not for the future of the project as a resource in a professional development network
17:00:45
Xach
makomo: i haven't tried cl-typesetting, but marc battyani always seemed to have really nice examples of its use.
17:00:46
jackdaniel
I don't understand the last sentence, but there is no need to elaborate, I've said everything I wanted to say ;)
17:01:59
jmercouris
An open source project is a basis for companies to work in a type of "open innovation network", several companies contribute to a core technology and leverage it to build their products. The value of an ncurses backend to building this network value for McClim is most likely very low, it basically only benefits yourself
17:03:36
jackdaniel
also, having such degenerated backend will benefit testing some hard-to-notice corner cases
17:03:54
stacksmith
jackdaniel: I am happy about the cl-charms backend - the x font rendering was hurting my eyes.
17:04:06
beach
jackdaniel: How dare you work on something that you won't make money from directly and immediately. That is so un-modern of you.
17:04:21
jmercouris
jackdaniel: It'll benefit the community, as some lisp developer will want to use the backend for sure, it will not benefit the project in a professional sense, because most likely no company will comission a curses backend for their software, and thus you won't win more support or sponsorship for the development of mcclim for the purposes of utilizing the curses backend
17:05:02
Shinmera
Professiona.L "engaged in a specified activity as one's main paid occupation rather than as an amateur.
17:05:06
jmercouris
beach: I work on plenty of projects that make no money, I'm not trying to shame someone for that, I don't think you're understanding me
17:05:45
jmercouris
We agree that you are a professional developer, but is everything you program something you program professionally?
17:05:49
dim
jackdaniel: you know (of) libcaca (coloured ascii-art) and its availability as a VLC backend?
17:06:37
_death
jmercouris: many companies have internal tools, I can imagine some of them have a text UI for a variety of purposes
17:06:45
jackdaniel
fun stuff indeed. I've also written a checkers game when I was at university in ncurses (with C)
17:07:10
jackdaniel
when you have scaled down the font (to something really small) it actually rendered circles :)
17:07:37
Shinmera
jmercouris: At the expense of others that don't want to see off-topic bullshit. Now drop it.
17:09:01
jmercouris
I'm trying to point out that more often than not the "#ontopic" discussion is used to silent dissent than anything else
17:09:57
beach
stacksmith: Several long-term participants have such privileges if that is what you mean.
17:12:29
dim
ignoring the group's rules and stuffing one own's wisdom into every bits of discussion, ignoring other participants altogether at best, but more often only what they say… we're damn near to a troll here, if you'd ask me
17:15:59
dim
so what about a turbovision backend for McCLIM and then an interactive development environment (editor, debugger, etc) all included there? ;-)
17:19:39
_death
I think clisp is the smallest thing that also fits "CL implementation".. and wasn't that 2-3 megs?
17:23:53
jackdaniel
you'd have to check by yourself (compiling it should be easy, it is one file) - don't forget to strip the binary
17:26:53
_death
jackdaniel: ok, so what's truly important is that floppy is still an option.. lisp has a better chance to survive the apocalypse