freenode/#lisp - IRC Chatlog
Search
15:00:51
Josh_2
Why would you want to do that when you can just write your own versions and not mess with the base image
15:26:44
aoeu256
I mean the macros return a new version of the function, this type of macro is an adverb. Like memoize, cached, printargs, except unlike memoize it can look at the source code of the function rather than treating the function like a "black box".
15:28:06
aoeu256
Instead of writing a new version of defun, it takes an old function applies a macro to its source code and returns a new function using that source code.
18:27:57
Fade
I've been playing with Sly after having used Slime for years. Are there any Sly experts in the house?
19:29:37
jackdaniel
I'm not sure if it is a kind of grateful response I've expected for caring to answer
19:33:41
Harag
I have thrown the kitchen sink at all the limits and for a simple task it still crashes
19:35:07
jackdaniel
if it still crashes from develop branch please report an issue, if you want to build it it is as easy as: ./configure && make -j5 && make install
19:37:03
jackdaniel
and if you have enough of ecl, you may try ccl which is more often recommended than ecl for general purpose programming
19:41:15
jackdaniel
btw if you have safety 0 specified somewhere I wouldn't be surprised if ecl had miscompiled things
19:49:49
Josh_2
If I want multiple sites to run off one hunchentoot server, where they can share the same urls like "home" "login" etc I have to split things up with packages?
19:57:54
jackdaniel
I would still advise using develop. we didn't make a release for 3y (we hope to make one this year), so it means that this branch has 3y of bugfixes and improvements in it
20:08:06
alandipert
does LET unintern the names of bindings unless they were interned already? Or is something else going on there
20:12:29
jackdaniel
in older versions some cli arguments were specified with a single dash, like -heap-size
20:13:40
alandipert
p_l my understanding is that symbols that are not qualified are interned when they are ready, so I would imagine when let sees them they have been interned already
20:14:59
p_l
alandipert: yes, but technically they don't have to be - LET-specific associations for those symbols exist only at compile time
21:00:41
jackdaniel
I would suspect that storage-exhausted is a result of too little free disk space
21:16:28
Harag
:jackdaniel ...and I have 32gb of ram at leaste 27 of that is available if I can just figure out how to get ecl to use it
21:18:07
Harag
:jackdaniel the fact that it is crashing on just quicloading a project means that --heap-size is not the right parameter or its being ignored
21:18:57
Josh_2
Not that important, but why have 5 lisp images hosting 5 sites when I can host 5 sites on one image
21:22:11
Harag
:jackdaniel ...I downloaded ccl and that at least loads the project but its stack is also crashing when I run tests and it has --heap-reserve and thats also not making any difference
21:24:16
Harag
:Josh_2 ...why the 5 lisp images? 5 different projects/packages with 5 different ports yes but not 5 images?
21:25:11
aeth
ooooh, using 32 GB RAM in CL... that's an interesting idea. Maybe I should do a really large prime sieve.
21:26:26
Harag
:aeth .. the test I am trying to run will use maybe 200 mega bytes but even the big tests dont go over 12gb in sbcl
21:27:09
Harag
:aeth ... I have a commercial site that runs at 7gb used and its been running like that for 10 years
21:27:13
Josh_2
Harag: why would you want someone who compromises one site to effect all the others?
21:30:34
Harag
:Josh_2 you run a hunchentoot server for each site so one site crashing is not going to effect the others
21:36:38
Harag
:jackdaniel ...ecl crashing on loading my project reports ...Broken at SI:BYTECODES. [Evaluation of: (ASDF/PARSE-DEFSYSTEM:DEFSYSTEM "ironclad"
21:39:01
Harag
any one out there feel like running a test for my project on a working lisp other than sbcl... please
22:44:21
Harag
:jackdaniel ... got ecl and my project working together...I deleted .cache...rebooted and started ecl without --heap-size
22:47:17
Harag
:jackdaniel ...I have no idea what the real issue was ... maybe it was just a tired user...thanx for trying to help
22:51:56
Harag
:jackdaniel ... specifying --heap-size still causes the crashes... should it be specified in bytes, kb, mg or gb?
22:58:51
Harag
how should i handle version numbers for quicklisp submission? Should I bump it each time there are any changes (bugs or enhancements) to my project or just when major milestones are met?
23:04:12
aeth
I don't think versions are even required for Quicklisp. There doesn't seem to be a widespread standard like semver in the node.js community.
23:05:48
no-defun-allowed
pretty sure ASDF likes some kind of semver-like numbering scheme for defsystem :version
23:31:49
aeth
no-defun-allowed: semver is just the only reasonable, standard thing you can put in a version string other than the old-Linux-style version number scheme that even the Linux kernel doesn't follow anymore because it means you could be stuck on 2.x forever
23:32:42
aeth
I suppose you could also do something like semver but have a "supermajor" in front for seriously big releases or something. I'd actually prefer this, mainly to avoid seeing "27.x.y"
23:33:28
aeth
but that'll probably never be widespread so I've given up on clever versions and so yyyy.mm.dd.patch is probably the way to go
23:34:21
Harag
ok well I am at 0.0.2 and it wont move from there for some time took 2 years to go from 0.0.1 to 0.0.2 ;)
23:38:12
aeth
Harag: exactly why just dotifying the ISO 8601 date format and then having an optional patch at the end is probably the way to go, especially if you don't even release multiple times a day
23:40:56
Harag
so 2019.08.19 good enough? ... I am going to forget to update the bloody thing... lol
23:41:59
aeth
some software's still in use from the last century, and some will probably still be in use in the next one
23:46:59
Harag
so the project has many sub projects...each one extending functionality with their own .asd's should all of those just stay in synq version wise or should they reflect only changes them them?
23:48:58
aeth
this is the beauty of time-based version numbers. If A has a release 2019.07.14 and B has a release 2019.07.22 and then A has a release 2019.07.27 and B depends on A then you know for sure (assuming things were tested correctly) that B should work with the 2019.07.14 release even if there's a regression in 2019.07.27 that breaks it.
23:50:13
aeth
i.e. any package should work on the largest version number of a dependency package that is less than its own version number, assuming that both packages use the same system
23:51:30
aeth
(less than its own version number, or equal to its version number except in the patch version, which could be higher, if they're both released on the same day)
23:53:32
aeth
Now, this might not always happen because you could be developing based on the updated-once-monthly Quicklisp version of the library, but if that doesn't happen, that's what the patch version is for, which doesn't have to be a same-day fix of that release.
23:55:48
Harag
the project is used by other "live" projects already so, development is as and when needed. The only reason then project code changed a lot in the last week was of refactoring and finally completed functionality... as well as making it play nice with other lisps etc for quicklisp's sake
23:57:10
Harag
there is one section that still needs some tlc code wise and one more feature then this project should come to rest mode
23:59:40
aeth
Oh, in anticipation of objections hours later (because this is IRC), yes, time-based versions don't have to work as well as I just described... but in a world where the libraries are fairly stable, like with CL, it probably would look like that.
0:00:27
Harag
well thats that... version numbers updated quicklisp issue created... now I can get some sleep