Search
Wednesday, 13th of September 2017, 7:55:54 UTC
8:55:42
Posterdati
so slime 2.20 has got problem with ecl :( "14 is not a valid index into the object "_""
8:56:07
jackdaniel
are you sure you don't have some malfunctioning cache?
8:56:17
jackdaniel
rm -r ~/.slime ~/.cache/common-lisp
8:56:19
jackdaniel
may help with that
8:56:30
Posterdati
probably, let me check
8:56:31
jackdaniel
because it looks like you have incompatible FAS file
8:57:04
Posterdati
ecl works fine from bash, so it is a cache or slime problem
8:58:47
Posterdati
always forgot the .cache stuff!
8:59:25
Posterdati
jackdaniel: do you think it is better to use the original asdf instead of the one from ecl?
9:00:14
jackdaniel
if you need something from upstream ASDF, then you should use it, but mind that it breaks backward compatibility from time to time
9:01:48
Posterdati
I'd like something rock solid so I can distribute the complete system on an sd card
9:05:33
Posterdati
problem solved... seems so!
9:06:57
Posterdati
jackdaniel: thanks
9:11:29
Posterdati
jackdaniel: after all you can say that ecl entered in the distribution power system :)
9:13:20
Posterdati
jackdaniel: the project I presented in ecl quarterly evolved
11:34:13
Posterdati
jackdaniel: anyway seems that clearing .cache did not solve the problem: "14 is not a valid index into the object "_""
11:36:11
jackdaniel
you have some old "fas" file from somewhere (compiled with different ECL version)
11:36:16
jackdaniel
that's the only explanation I can think of
11:37:00
Posterdati
in the projects directory?
11:37:11
jackdaniel
I don't know, loading *which* file causes this error?
11:37:20
jackdaniel
you should have that on your backtrace I suppose
11:37:51
Posterdati
the error occurs randomly while I'm running slime
11:38:36
jackdaniel
no idea where it comes from, I need more information for this
11:38:51
jackdaniel
I will check with new slime later this week
11:48:22
ralt
Posterdati: is that with jsown?
11:49:07
ralt
that sounds a lot like https://github.com/madnificent/jsown/issues/18
11:49:26
ralt
jackdaniel: it's the INCF THE bug ^
11:50:41
jackdaniel
but is this really what Posterdati has?
11:50:49
ralt
it's the same error message
11:51:15
ralt
which is just a consequence of the incf the bug
11:51:50
jackdaniel
Posterdati: this bug is fixed in ECL's devleop, not present in any release yet
11:51:57
ralt
it might not be the issue, but that does ring some bells
11:52:39
ralt
I wasn't there at the beginning, so I don't know if it's using jsown
11:52:53
ralt
(I don't know the context of the discussion, that is)
11:57:05
ralt
jackdaniel: I was just saying that if it's using jsown, it's the INCF THE bug
11:57:27
jackdaniel
right, I understand, thanks
11:57:40
jackdaniel
Posterdati: could you check with develop version if the bug disappears?
11:58:00
jackdaniel
also pasting backtrace for such error would be awesome
11:58:07
jackdaniel
for instance on paste.lisp.org
12:14:55
ralt
jackdaniel: would it be a good/bad idea to have a `#+ecl2` feature or something? something that would let me differentiate whether this bug is going to occur or not.
12:16:24
ralt
not sure it matters though
12:17:31
ralt
it's just for this jsown lib...
12:17:43
jackdaniel
ECL has better mechanism form that: ext:+ecl-version-number+
12:19:27
jackdaniel
#+ecl #.(when (>= ext:+ecl-version-number+ 160103) '(list 1 2 3))
12:19:41
jackdaniel
in your case (> ext:+ecl-version-number+ 160103)
12:21:31
jackdaniel
or <=, I don't know your exact case
12:32:31
ralt
jackdaniel: the use case is there: https://github.com/madnificent/jsown/pull/19
12:43:17
jackdaniel
#+ecl #.(if (> ext:+ecl-version-number+ 160103) `(the fixnum …) `…)
13:15:14
ralt
Posterdati: are you using jsown?
13:15:28
ralt
one of your dependencies maybe?
13:15:46
Posterdati
I'm using iolib mainly
13:21:12
Posterdati
http://paste.lisp.org/+7MHK
13:24:04
Posterdati
http://paste.lisp.org/+7MHK/1
13:24:25
jackdaniel
it might be that adding threading to ecl's swank broke something too
13:24:32
jackdaniel
I'll have to investigate that further
13:25:01
Posterdati
this arose after slime update to 2.20
13:25:26
jackdaniel
yes, you've alrady mentioned that
13:25:30
Posterdati
it seems to happen sporadically at random time during slime session
13:25:47
Posterdati
no particular operation
13:28:34
Posterdati
I'm monitoring *slime-events*
13:28:39
Posterdati
no particular problem
13:30:42
ralt
Posterdati: could you try with the develop branch of ECL?
13:35:16
ralt
Posterdati: git clone https://gitlab.com/embeddable-common-lisp/ecl && cd ecl && ./configure && make && sudo make install
13:35:41
ralt
git checkout develop before "./configure"
13:35:59
Posterdati
http://paste.lisp.org/+7MHM
13:36:16
Posterdati
the error came out while I'm navigating my source
13:40:44
Posterdati
ralt: I'm already using that!
13:46:04
Posterdati
could be the slime-fancy or slime-xyz code that fails
13:51:11
Posterdati
I'm testing the result
13:55:16
Posterdati
(setf children-of) is not a valid generic function name
13:55:54
Posterdati
[Condition of type SI:SIMPLE-PROGRAM-ERROR]
13:56:01
Posterdati
during slime loading
14:06:21
Posterdati
is the swank-trace-dialog.lisp
14:46:46
Posterdati
my ecl version was too old
14:47:00
Posterdati
I pulled the last git and it works now
14:55:23
jackdaniel
no issue I have to investigate?
14:56:04
Posterdati
now I'm quickloading my project
14:56:14
Posterdati
but no error during slime startup
14:56:34
Posterdati
now I'll check editing the source
15:00:20
Posterdati
the error come back!
15:01:40
jackdaniel
so I probably made a mistake when I made a pull request to slime project enabling :spawn communication style
15:02:19
Posterdati
the error is the same
15:11:22
Posterdati
http://paste.lisp.org/display/355736#1
15:11:42
Posterdati
the integer number is variable :)
15:33:56
Posterdati
it is related to swank autodoc
15:38:04
Posterdati
in fact the error come up when cursor goes over symbol/function names
15:41:20
Posterdati
jackdaniel: are you pushing a new version?
15:41:39
jackdaniel
you mean new release of ECL?
15:41:59
jackdaniel
if you look at gitlab, there is 16.2.0 milestone
15:42:04
jackdaniel
we still have a lot of issues to solve
15:42:14
jackdaniel
but we have advanced a lot already
15:42:34
jackdaniel
maybe I'll need to review some issues which I want to push even further in time
15:42:50
Posterdati
alas this last issue is not critical it is annoying
15:44:05
jackdaniel
Posterdati: could you describe it on gitlab? so I won't forget it
15:50:02
Posterdati
are we sure is it ecl?
15:50:21
Posterdati
I mean could be slime...
15:50:25
jackdaniel
it's most probably swank-ecl, but put it in ecl
15:50:40
jackdaniel
I will fix it in slime, but it will be easier for me to remember
15:50:44
jackdaniel
if you put it on ecl's tracker
15:51:16
jackdaniel
if you could introduce minimum example to reproduce the issue that would be even greater
16:05:51
Posterdati
https://gitlab.com/embeddable-common-lisp/ecl/issues/406
Wednesday, 13th of September 2017, 19:55:54 UTC