freenode/#lisp - IRC Chatlog
Search
10:43:15
minion
why Lisp: The first chapter of Practical Common Lisp: http://www.gigamonkeys.com/book/introduction-why-lisp.html
10:46:47
Younder
I take it from flip214 that I need to speak perfectly conceptually clear. And I rely on the spell checker because a simple comma can label me as an idiot,'
10:47:59
Younder
I take it from flip214 that I need to speak perfectly conceptually clear. And if I rely on the spell checker because of a simple comma can label you can still me as an idiot.
11:12:28
myrkraverk
The colon in question is inside " "s, so I really don't understand the error message.
11:20:42
Younder
Whatever, on a remote terminal, vim with syntasic is just about the best you can get.
11:26:42
Shinmera
Tramp is basically just an mode for emacs that copies the file in the back through SSH when you save/load.
11:34:44
solene
I'm displaying text in a terminal and I would like to continue scrolling using #\Space
11:38:32
myrkraverk
When I do (coerce 'simple-string foo) ; where foo is (vector character 40) why do I get: SIMPLE-BASE-STRING can't be converted to type (VECTOR CHARACTER 40) ?
11:40:01
jackdaniel
also, simple-base-string has base-char elements, while there is no guarantee, that character is a base-char
11:40:31
myrkraverk
The function I'm trying to feed this accepts simple-string and simple-base-string.
11:45:11
Younder
you might wanna look at: https://github.com/sbcl/sbcl/blob/master/src/code/reader.lisp
11:47:27
myrkraverk
Such as custom looping through the vector to create a simple-string, or something.
11:52:21
pjb
solene: if you insist on SPC, have a look at implementation specific features such as #+clisp ext:with-keyboard, or at cl-charms (ncurses).
11:52:32
solene
pjb, I would like to be able to use both, I use get-char now and it works well with RET (it's also possible to type others characters in my software)
11:52:55
solene
I think I'm going with a little of C code, I already have an external C library so most of the work is already there
11:54:04
pjb
You may also use termios, with cffi, but I'd be weary of changing the characteristics of the file descriptor under CL streams.
12:24:27
jackdaniel
pjb: cffi = calling foreign code, that's all. You don't have to touch C to use functions using this calling convention
13:04:48
pjb
jackdaniel: cffi has all the disadvantage of the C code compiled with the C compilers requiring cffi.
13:51:59
logicmoo
macro-function lambda lists are always (form env) .. but not (&whole form env) .. right?
13:54:52
pjb
It then uses destructuring-bind to split it into the macro lambda list (minus &env which is processed by defmacro and passed as the env argument).
13:56:45
logicmoo
this leads me to another question (barely related) nut since you mentioned the destructing-bind was wondering if documentation strings would be left over inside the stored form
13:58:27
pjb
destructuring-bind doesn't take a docstring, but it take declarations. Since declarations may occur after the docstring, it would break if it was left.
13:59:46
pjb
Notice that it all relies on defmacro, which is what is specified. Your defmacro could very well generate the macro-function differently, not using destructuring-bind.
14:00:26
pjb
But if you assume it, then this gives constraints on defmacro, such as the handling of &env, of the docstring, and of any declaration concerning the &env parameter.
14:02:17
logicmoo
i neglected to also synthsize the macro-function though so making sure its proper as one might expect to (macro-function 'symbol)
14:02:41
pjb
(defmacro foo (abc &env *env*) (declare (integer abc) (dynamic-extend *env*) (special *env*)) (foo* abc (g env)))
15:07:42
drmeister
Hey lispers - Xach told me a few weeks ago about a mechanism in quicklisp to bundle systems for deployment. I'm trying to jog my memory - does anyone know about a quicklisp facility for bundling systems?
15:13:45
drmeister
Ok - that's ok - my end goal is to construct a monolithic fasl that contains code from several quicklisp systems, several quicklisp systems that are in my local-projects and several asdf systems that are provided by my source code.
15:15:14
drmeister
Along the way I need a list of source files that will serve as dependencies for the build system that will build this monolithic fasl.
15:17:26
drmeister
Currently I use (ql:quickload "main-system-with-many-dependencies") to find all of the asdf .asd files in quicklisp/local-projects and quicklisp/dists/quicklisp/software so that I can grovel the "main-system-with-many-dependencies" to build this list of common lisp files that will be the dependencies of the monolithic fasl.
15:19:46
drmeister
Is there a way to get quicklisp to find and load all of the .asd files but not compile everything?
15:24:22
drmeister
Sorry if this sounds a bit confused - I'm a bit off the map here - trying to figure out how to properly merge building a monolithic fasl file with our 'waf' build system (written in Python) and get waf to know the dependencies of the monolithic fasl file.
15:24:28
dmiles
is it acceptable practice to store important parts of the implementation things in the symbol-plist or should it be kept clear for users?
15:28:38
dmiles
implementation package makes sense .. i was thinking about though as i am loading a file.. for things like warnings
15:36:23
dmiles
it transplies the bodys to prolog.. then uses the internal compiler of prolog to make tramploines
15:38:32
dmiles
but i also expect macroredefinations and want ot keep there sources arround so i can recmpile as needed
15:40:41
dmiles
for exmaple my symbol-function accessor lets them supply an env so they can peer into the flets
15:41:29
thebigj
Friends, I am trying to understand https://dpaste.de/ZRvU/raw this sample common lisp code given as the example in Aspected Oriented Programming ECOOP'97
15:41:54
thebigj
From https://en.wikipedia.org/wiki/Common_Lisp#Syntax I can understand first line is defining a function named or!
15:42:29
beach
thebigj: How come you need to understand Common Lisp code if you hve no experience with it?
15:43:24
thebigj
beach: I have experience with programming. If someone can explain in words I will be able to catch them and understand what the code does.
15:45:01
beach
If the image A has a non-nil pixel in it, then that pixel is copied into the image RESULT, otherwise the pixel in image B is copied.
15:45:45
beach
The code is also badly indented, and the naming is bad. We don't use the SET- and GET- prefixes much.
15:46:02
dmiles
btw i finnaly letting the compiler write the functions for me .. (instead of by hand).. here is one it write: https://github.com/TeamSPoon/wam_common_lisp/blob/master/prolog/wam_cl/places.pl#L20-L52
15:48:23
beach
thebigj: I still don't understand why you need to understand Common Lisp code when you have no experience with it. Are you trying to learn Common Lisp? If so, it is probably better to read a book.
15:51:56
_death
OR is a weird operator to use in that context, maybe it should be LOGIOR.. and then, you'd want to generalize it so that it'll be (combine image1 image2 #'logior)
15:52:52
_death
and then you'd likely want to be able to pass a result image rather than creating a new one
15:53:18
Bike
Right. Well, the code is ORing two images together in an explicit nested loop like beach said. It's not great code as everybody said, but that's the gist
15:59:15
dmiles
beach: i dont suggest trying it now but here is what was working a few days ago: https://github.com/TeamSPoon/wam_common_lisp/blob/master/README.md#howtoquickstart
16:30:42
beach
dmiles: I do not use Prolog, so I'll be interested in your implementation only if it has some advantages over what I currently use. Sorry about that.
16:32:50
beach
dmiles: Interested in using it, that is. I might be interested in implementation strategies.
16:35:40
dmiles
i used a prolog atom (it is a special interned object that acts like aprimary key in the database) .. i record its superclasses and that is it. i lazily create prolog fields in the 'asserta database' as they are either first read or set
16:38:40
dmiles
if someone adds a cons to a field.. i have to create an area that is tracked and modifiable
16:38:42
beach
By "standard instance" I meant an instance of a standard class, and not necessarily a class.
16:39:26
dmiles
i meant an instnace .. except i should have not siad superclass.. i should have said one of more classes it is an instance of
16:40:47
beach
But, again, why does it matter what you store in a slot? Why is storing a CONS different from storing, say, a symbol?
16:43:45
beach
I am sorry, this is not working. For every question I ask, your answer introduces at least one new term I don't understand.
16:44:00
dmiles
(let ((x (cons 1 2))) (setf (slot-value ....) x)) that slot gets created on first use .. in live memory
16:46:22
dmiles
until a the slot is needed .. at which time i confirm on the class metaobject the slot is even valid
16:47:52
beach
I still don't understand why there is a difference in what happens in the standard object (call it obj) when you do (let ((x (cons 1 2))) (setf (slot-value obj ...) x)) compared to (let ((x 'a)) (setf (slot-value obj ...) x))
16:47:54
dmiles
this though wasnt a premature optimazion (though it would sound like one) it is that it is to allow a non determistic deductive system
16:49:20
dmiles
whay it maters if it is a cons or array vs a constant because constants are all identically idetified
16:50:27
beach
And, how are you using the word "constant" here? Certainly not the way it is used in Common Lisp.
16:52:10
beach
(let ((x (cons 1 2))) (setf (slot-value obj ...) x) (eq x (slot-value obj ...))) => True
16:55:27
beach
OK, let's do this instead (defun ff (x) (setf (slot-value obj ...) x)). Does it matter what type X is what happens in the (setf slot-value)?
16:55:48
dmiles
Objects are all inert pictures as well.. so to put something modifiable in it.. i have to make sure i can modify it later without modifying the object
16:58:30
beach
So, suppose the function above was called like this one time (ff (cons 1 2)) and like this another time (ff '(1 . 2)).
16:59:46
beach
The result of evaluating (cons 1 2) is mutable, but the result of evaluating '(1 . 2) is not.
17:00:36
beach
About half an hour I asked whether the (setf slot-value) behaves differently if it is a CONS and if it is a symbol, and you said yes.
17:01:20
beach
I don't think I'll ever understand what you are saying, so for now, I have to give up. Sorry.
17:02:13
beach
You might want to make sure you use standard terminology if you want people to understand what you are doing.
17:05:16
dmiles
worse yet, i try (though not even doing that good a job of it) trying to use more C terminology than prolgo since people have more experience in talking about Lisp in C languages
17:06:54
dmiles
as far as what my implementation does in general is acts very lazy so i pay penalities at JIT compiler time
17:10:58
dmiles
there are many optizations that work on the warren abstract machine (wam-cl) that dont work so great on conventional machines .. the warren abstract machine was created to do things lisp coders where doing at the time with expert system
17:14:04
Bike
it seems like you're talking about basic semantics rather than optimizations. and like, "levels", "neutral", none of those are lisp terms or c terms.
17:17:20
dmiles
setf after its ben resolved to call something, will call something that will have several levles
17:21:24
dmiles
Bike: if it creates a frame than sure i suppose it could be a a differnt level especially if we have differnt shadow bindings in each c1 (even if TC)
17:22:55
dmiles
a shadow binding is where a dynamic symbol has a value that changes durring child frames
17:33:53
dmiles
the problem with the term lexical binding is it is used to not biding both dynamic and non-dynamic variables
17:38:36
beach
drmeister: You REALLY need to make sure you use standard terminology, or else you won't be understood. And people might lose interest, the way I did.
17:39:53
dmiles
i doubt my answering of questions will be helpfull as i may not be able to answer them in the terminaolgy desired
17:41:41
beach
dmiles: But, you see, if you don't even know the terminology of Common Lisp semantics, your discourse will make some of us doubt that you are capable of implementing the language at all.
17:46:14
beach
I think it has to do with my training as an engineer, where every word may have some technical meaning.
17:53:59
dmiles
when answering questions about how i instance a standard object, i assumed it is becasue it was hard to think aobut in a language that doesnt have mutability or objects being able to do so, so i was trying to answer why it wasnt impossible to do in prolog.. thus the hacks and inventions
18:00:13
beach
It is safest to assume that the people you are talking to here have very little knowledge about Prolog, especially modern implementations of it.
18:13:00
stylewarning
Hey all, is there documentation anywhere about statically linked C libraries with SBCL/CFFI?
18:25:15
stylewarning
This is so vague to me: https://common-lisp.net/project/cffi/manual/cffi-manual.html#Static-Linking
18:32:19
stylewarning
I see slides, I see reference in the CFFI manual. Of course, in usual Lisp tradition, no comprehensive docs, no examples, no warnings, etc.
18:36:26
Shinmera
If you people knew just how much undocumented, incomprehensible source was out there in other languages...
18:45:40
stylewarning
When I use static-program-op, I get some business about "undefined symbol _main"
18:50:14
stylewarning
Found Fare's static example: https://github.com/fare/workout-timer/blob/master/workout-timer.asd
19:04:45
jeosol
CLOS design question here. What is the better way (for maintainability) to combine two methods defined on two classes A and B, with B inheriting from A. I normally add :before qualifiers to the method with A's object and then just call the method on B, which is the primary method. I have cases where the defmethod on A is called in B using (call-next-method) (this does not use :before tag on defmethod on B).
19:05:57
jeosol
typo correction: in the latter, I mean't (this does not use :before qualifier on defmethod on A
19:07:20
jeosol
I am asking to the general forum. It's more of a design consideration. I normally use the option adding :before to the defmethod on the base class, but have seen codes with call-next-method obviously to handle some specially cases down the line
19:08:09
beach
I don't think there is a general answer to your question. The right answer has to do with the protocol that you are designing. Specifically how you describe the role of the primary method(s).
19:10:06
rpg
Also, notice that :BEFORE methods are really only done for side-effects, since they don't change what the primary method computes.
19:10:41
rpg
... and they can't establish a context for computation necessarily because they are not scoped *around* the primary method.
19:10:57
jeosol
In that case, in my case, the method on the base class only sets up things for A, so I think it's better to use :before qualifier and rely on the standard method combination
19:11:23
jeosol
Yes, I have read that book. I also recently read Andreas compilation and have seen different paradigms used
19:11:45
beach
jeosol: I for one find it hard to find typical use cases from such a general description. But you are probably right.
19:12:22
jeosol
I do understand how CLOS works, I am just concerned about what makes for better code down the line if I get another developer. The organization of class hierarchy is in my head and the code base is getting fairly large
19:13:13
jeosol
thanks for the comments guys, I think it's better to use the :before qualifier on the base class.
19:13:21
beach
I understand, and (again) I think the answer depends on the real meaning of the operations in your protocol, and that no general answer exists for all the cases that fit your description.
19:14:25
jeosol
The book I was referring to was Objected Oriented Programming: The Clos Perspective Edited by Andreas Paepcke
19:18:07
jeosol
Chapters 1, and 2 are very good, respectively Introduction to CLOS and CLOS in Context.
19:19:41
stylewarning
rpg: Do you have some insight as to what this operation actually allows one to do: https://common-lisp.net/project/cffi/manual/html_node/Static-Linking.html
19:20:29
jeosol
Apologies for not being specific enough. Those are the names of the first two chapters.
19:20:53
rme
I always say "Keene", because I can't ever remember if it's "Sonya" (in the text and on the copyright page), "Sonja" (on the cover), or "Sonia" (just another spelling).
19:23:54
jeosol
I think so too, it's a great book. I like it a lot especially the application of CLOS in real applications.
19:25:19
rpg
stylewarning: Not as well as I should: I don't really use the bundle operations myself, and they were added to ASDF without much participation on my part. One thing I do know, though, is that moving images for some lisp implementations from one location to another can fail as they encounter local dynamic libraries that differ from what they "expect."
19:27:29
rpg
stylewarning: But to be honest, I don't know if this problem can occur even without FFI -- I.e., we can dump an image onto a lisp that is sitting on an operating system that has different libraries, providing different entry points to the lisp implementation itself.
19:28:43
rpg
I have decidedly mixed feeling about ASDF's mission-creep from loading systems to providing delivery solution. I believe ASDF kind of slipped into this as it tried to accommodate "compile-to-C" approaches to CL (notably ECL).
19:29:19
stylewarning
s what it seems. Then again, a robust and flexible solution across all implementations seems nice.
19:29:57
stylewarning
I myself am going through a lot of effort right now to make a command-line program faster to start/stop, and it seems like a major bottleneck is the dlopen-ing of 4 libraries.
19:30:51
rpg
stylewarning: So this is for working with FFI? Or is it just a slowdown when the lisp implementation itself is starting?
19:34:06
rpg
stylewarning: then this operation might work for you. You might well get a more portable application out of it, too.
19:34:53
stylewarning
I'm fighting to get it to work. I'm getting some issue where it says it can't find _main when doing some linker operation with "static-runtime_XXXX"
19:44:41
rpg
stylewarning: I'm afraid I don't know that much about the internals of SBCL. But this looks like a problem with CFFI, I think. Not sure why it's looking for main, or who's doing it. Do you have some kind of backtrace for this?
19:53:24
rpg
are you sure that "make" maps to the operation you expect? Looks like it should. But just in case -- you're also testing with (oos 'static-program-op "static-fire/static")
19:54:11
stylewarning
rpg: I'm fairly certain make is doing what it should because it's calling the linker. But yes, in that case, using oos results in the same failure.
19:55:19
stylewarning
I poked around in the source and it seemed like MacOS was supported (mostly through the #+bsd feature)
19:58:14
stylewarning
rpg: Further hint: On Linux, at least `make` returned T; not sure if/where it wrote to.
19:58:18
rpg
stylewarning: I just don't know what compilers support compiling to .o instead of .dylib on that platform.
19:59:21
rpg
stylewarning: on linux you should go and root around in the .cache to see if you can find the compiled output. There's some way to redirect the output, but it's not very tasty (IMO), again because of the extension from fasl-building to executable-building.
20:06:08
fourier
whats the easiest way to compare 2 class instances which has only numeric slots without manually comparing slots ? iterate over slots using MOP and compare them?
20:14:45
stylewarning
Is it too hard to write (with-slots ... (with slots ... (and (= a-1 b-1) (= a-2 b-2) ...)))?
20:19:28
shka
fourier: if those are just numbers, stick all of those into vector and use vector= from alexandria
20:20:09
Bike
It's easy. You have (defclass foo ...). You write (defun foo= (foo1 foo2) (and (= (foo-a foo1) (foo-a foo2)) (= (foo-b foo1) (foo-b foo2)) ...))
20:20:47
fourier
problem is what i dont know accessors/fields. the struct is generated by cffi's define-c-struct-wrapper
20:22:43
Bike
I mean, defcstruct doesn't even make a lisp class you instantiate lisp objects of, does it?
20:29:58
Bike
This seems to define a CL class that mirrors a C struct and can be initialized with one (but thereafter loses any connection to any C object)
20:33:06
fourier
another question though. is elsconf mailing list dead now or I'm not properly subscribed (resubbed with another email) ?
20:38:54
rumbler31
what was that thing about specifying symbols that are not interned? like if you wanted to ql a package and you didn't want the name you provided to be interned as a symbol. I'm not sure what to google for that
20:39:26
Bike
though i think for quicklisp they're string designators, so you could just throw it a string.
20:40:40
rme
fourier: I don't think there's been any traffic on the ELS mailing list. According to the list archives, the most recent message was sent on November 29th.
20:43:48
Bike
rumbler31: i think quickload is a normal function, so it evaluates its argument (and #:quickproject is an unbound symbol)
20:44:42
rumbler31
I could have sworn that this #: was discussed in the context of quickloading before but
20:46:20
rumbler31
so the reader reads #:quickproject, doesn't intern it, it gets passed to quickload as (quote quickproject)
20:55:54
Bike
it's nothing special. it's like if you wrote (ql:quickload oiwjefoansd). oiwjefoansd is unbound, so no good
21:28:47
aeth
If you don't need performance, use s-expressions. If you need performance, nothing beats a custom binary format.
21:29:45
pjb
You may want: (with-standard-io-syntax (let ((*read-eval* nil)) (read socket))) for more security, but it would be better to authenticate and validate the data otherwise.
21:29:46
aeth
json is problematic because the popular libraries are not two-way, i.e. CL -> JSON -> CL could produce different CL than you started with
21:30:04
pjb
and because there are limitations that depends on the programming languages on the other side.
21:31:02
aeth
I would generally treat JSON or XML as an external-to-CL interface and use either s-exp or binary serialization for CL-to-CL
21:31:36
jasom
[] <=> #() ; true <=> T ; false <=> nil ; {} <=> (make-hash-table :test #'equal) ; nil <=> :null
21:31:51
jasom
[] <=> #() ; true <=> T ; false <=> nil ; {} <=> (make-hash-table :test #'equal) ; null <=> :null
21:32:29
aeth
The only problem is people want to use lists and that opens up the '() vs false issue.
21:33:24
aeth
jasom: That's not as big of an issue because you can have a prefix for plists where the car is a 'tag' and the cdr is a well-formed plist
21:34:00
jasom
aeth: and few libraries actually do this; they rather have () be null false empty-array and empty-object
21:34:47
jasom
aeth: to be fair, I would *not* have come up with that mapping before having used a bunch of other mappings for a while first
21:35:59
jasom
anyways I was hoping for something *slightly* higher level; like "Send this message and signal some condition if it doesn't get through" which is ... harder than it should be using just sockets.
21:37:22
aeth
Never use conspack for this. It's like the Python serialization format. Too powerful so you have to trust it to load it.
21:38:26
aeth
Similarly, don't just eval s-expressions as data, write a mini-eval that only evals what you expect to run
21:38:53
aeth
jasom: I'm not sure if cl-store is safe or if it also suffers from serializing too much of the language
21:39:43
jasom
aeth: this is interesting; conspack specifically lists dealing with untrusted peers as a goal, right?
21:40:37
Fare
it works for things that are declared with e.g. cffi-wrappers; but so far no one has added support for declaring and linking arbitrary C libraries (either statically or dynamically).
21:41:24
Fare
The heavy lifting was done, but the definition of social standards was too much for me -- I have 0 energy left to fight the CL community.
21:41:50
rpg
jasom: They do require you to know a little bit about what you are sending/receiving because of the aforementioned issue of the differences between JSON and CL about what's "truthy"\
21:42:31
aeth
jasom: What you want is a very limited serialization imo. Numerical types, very basic data structures, etc. I personally think attempting objects is too much. Just write a way to handle that manually, on both ends.
21:42:33
rpg
I think CL-JSON lets you specify whether you want values to be lists or booleans. That must be done in the library, since it's not a JSON thing.
21:43:13
jasom
rpg: I see 3 ways of dealing with json that make sense 1) using non-native data types (I think jonathan does this) 2) Using a reasonable bijective mapping like I proposed 3) Requiring a schema of some sort so that ambiguities can be resolved
21:43:58
rpg
jasom: CL-JSON, IIRC, uses CLOS to specialize the translations for specific situations. But I haven't used it in years, so I may be misremembering.
21:44:31
jasom
rpg: I have gotten cl-json to use the mapping I propose. I also got yason to do it. However, they both have poor defaults.
21:44:41
rpg
JSON is a weird standard, because JavaScript's notions of truth, falsehood, and equality are insane.
21:45:08
aeth
jasom: Writing a library with your mappings would be helpful for the CL community imo.
21:45:20
jasom
JSON is not that weird; many languages distinguish between an empty array, a literal false, 0, and null.
21:46:33
rpg
aeth: I think it might be better to simply write a wrapper around the existing libraries that applies a different set of defaults. On the general rule that we are proliferating more libraries than we (as a community) can maintain
21:46:54
aeth
JSON is nice for languages close to JS: Lua, Python, Ruby, Perl, etc. It's tougher when you're in a language that's farther away from the 'scripting languages' family.
21:47:14
Fare
rpg: and yes, the bundle support was born out of the old asdf-ecl + desire to match xcvb in terms of dumping executables.
21:47:46
jasom
back to the subject; I may use something like zeromq or nanomsg for messaging. Thoughts?
21:48:14
aeth
The languages I listed (mostly) have similar concepts to the subset of JS that's in JSON, e.g. Python "lists" and dictionaries.
21:50:17
jasom
jmercouris: most languages have more than one false value, rather than having a single value for all falsey things
21:50:20
rpg
For CL, you just need to have a sense of when the JSON field is of type boolean and when it's of type list. You kind of need that for, say C, as well, so I don't think it's a problem.
21:51:31
jasom
jmercouris: whether it's good or bad is up for debate, but JSON makes the assumption that "No Object" "False" and "Empty" are distinguishable
21:52:08
jasom
jmercouris: there are ergonomic values to having all of them being treated similarly, but you have less information when you cant distinguish them
21:53:08
jmercouris
I can't think of a scenario in which I've needed this distinction, but I'm sure there is one, something to think about I guess
21:57:23
jmercouris
Shinmera: One of the biggest java proponent arguments for all of this type safety bs, generics etc "Catch the problems during compilation time, save development time"
21:57:35
aeth
CL but statically typed would be the ideal programming language imo. Might need to separate '() and #f like Scheme, though
21:58:18
aeth
jmercouris: Shinmera iirc is working on a game engine. Game engines mostly deal with very well-defined types. Static typing is a huge win here.
21:58:43
Shinmera
The fact that Java sucks balls has very little to do with people finding advantages in strong typing.
21:58:57
jmercouris
Otherwise java programs should have considerably less development time over weakly typed languages, but they dont
21:59:12
jmercouris
Especially because of all of this compiler support and IDE support you get in java
21:59:37
aeth
Java sucks because it sets-in-stone 1990s C++ OOP fads while the world has moved on, even the C++ world.
22:00:16
jmercouris
there had been huge developments in JIT work at Sun, as well as Dtrace, and all sorts of facility integration that would have made java phenomenalj
22:00:31
jmercouris
I believe that Java could be so much more today if shit didn't hit the fan at sun
22:02:19
aeth
Idea? Yes. In practice? The JVM is designed for Java-like languages and the CLR is designed for C#-like languages, and actually-interesting languages can't run on either unless they were designed to make the compromises necessary to run on one or both.
22:03:10
jmercouris
aeth: Are we getting into this transpiler debate again? Last time I was disparaged as some insane heretic for suggesting the word transpiler exists :D
22:03:41
aeth
I think compiling to CL could support more interesting languages than compiling to the JVM.
22:03:58
phoe
In LOOP, can I somehow skip back to the beginning under some condition? Kind of like "continue" in C.
22:04:21
aeth
CL can recompile at runtime. Could you use this to make a JIT language that runs in a CL environment that's not JIT?
22:06:33
jasom
phoe: not really; you just need to have the rest of the loop under a continue, or put the body of the loop in a tagbody and GO to the end
22:07:06
aeth
jmercouris: I thought you needed two parts: (1) analyze the code and (2) recompile the code at run time. If #1 is sufficient, you could do #2 on a language that runs in CL
22:07:32
jmercouris
aeth: You're spot on with your definition, I'm just saying the purpose isn't to have a dynamic system, it's to make compilation quicker
22:08:04
Bike
plenty of jit systems take advantage to speed up generated code in ways that would be difficult ahead of time.
22:09:44
stylewarning
Fare: I have a little minimal example that I cannot get to work or that I do not understand.
22:09:55
Bike
you could do something dumb like (defmacro defunj (name lambda-list &body body) `(let ((counter 0)) (defun ,name ,lambda-list (when (> (incf counter) 10) (compile ',name)) ,@body))) and then you can define a function that's just interpreted at first and then compiled once it's run a lot.
22:10:55
jmercouris
have some sort of "probability" that some branch will be executed and compile or not compile different paths
22:11:19
jasom
I thought the main reason for JITs is to allow efficient compilation of languages where the types are mostly fixed at runtime, but mostly unknown at compile time
22:11:44
jmercouris
jasom: No, the types in java are always known at compile time, so there is your first counter example
22:12:59
dmiles
the reason for JVMs JIT i thought was because of limited memory, that it would be too much to compile the entire 9 million classes
22:13:02
Bike
you can do polymorphism without jit. it just makes it faster, hopefully. which is what jasom said.
22:13:03
aeth
jmercouris: I was more thinking about something like a modern JS-in-CL, which probably requires a fake JIT to support all the magic necessary to have somewhat competitive performance.
22:13:15
Bike
i suppose this means beach's gf dispatch business is jit compilation. that's kind of neat.
22:13:17
aeth
JS is a language that simply does not give enough information to produce fast code without a JIT afaik
22:14:35
dmiles
(the entire JDK and all the classes are "interpreted bytecode" ... some a few other langauges can be that way as well .. even lisp)
22:14:37
Bike
doing it for methods themselves would be a little trickier, you'd need to save method body sources, or at least enough of a source.
22:15:26
aeth
jmercouris: It's not just types, though, JS JITs take quadratic code and make it not quadratic. https://accidentallyquadratic.tumblr.com/post/142387131042/nodejs-left-pad
22:16:00
aeth
JS JITs are what you get when you take very bright minds and put them to work polishing a turd instead of creating the next generation of very high level declarative languages.
22:16:56
jmercouris
I have a lot of complaints about java, but it isn't the worst language in the world, it is very consistent for one, very structured
22:17:11
Bike
with beach's dispatch, if you call a gf with a combination of classes it's never been called with before, it compiles a new discriminating function.
22:17:14
jasom
this is why ltk is way faster to startup if you create and teardown your GUI before saving your image btw
22:17:25
jmercouris
Many JS apologists say "have you tried ES6", and the answer is no, nor do I care to
22:17:56
stylewarning
Fare: I was trying to compile an application, and I thought static-program-op might do magic to make dynamically loaded libraries load faster. I suspect this isn't true (that would be too magical).
22:18:05
jasom
Bike: oh, I missed that part; I was just aware of the part where it used first-class global environments to inline the GF dispatch function and recompile the usage points at redefinition time.
22:18:18
jmercouris
I don't understand how javascript isn't somehow a very blatant copyright infringement