greedom has quit [Remote host closed the connection]
lritter has joined #pypy
luckydonald has quit [Ping timeout: 256 seconds]
luckydonald has joined #pypy
<marmoute>
mattip: roundup was mostly ancient ^^
<mattip>
yup, but the absolute dominance of a single solution is worrisome
<LarstiQ>
Gitlab and Bitbucket in my mind are the main github competition
<LarstiQ>
although their models are of course pretty similar
<mgorny>
in general, any kind of SaSS for a major project is worrisome
<mgorny>
bitbucket is practically dead (by suicide)
<mgorny>
one account with sufficient privileges is cracked and the attacker can totally mess up your project
<mgorny>
(that's about any sass, not bb)
<mgorny>
and believe me, asking gh support to restore stuff from backup is well...
<LarstiQ>
mgorny: I see Bitbucket used a _lot_ in companies
<mgorny>
let me put it this way: at one point we've lost all pull requests from our repo
<mgorny>
and github couldn't restore them
<mgorny>
LarstiQ: well, i don't know about companies. i've mostly seen bitbucket used by people who wanted to use mercurial. then they removed mercurial support, and people were like "if we have to switch to git, we may as well use github instead"
Guest96 has joined #pypy
Guest96 has quit [Client Quit]
<krono>
code generation is a headache when Math 101 was 15 years ago
<krono>
cfbolz: do you remember small_list?
<cfbolz>
Vaguely
<krono>
generating storage classes that uses fields instead of lists…
<krono>
it has a tiiiny bit of automatic unboxing, which i need a bit more generalized
<krono>
anyway, maybe I can get it to a point where it can be a "rsmall_list"
reneeontheweb has quit [Quit: Client closed]
<cfbolz>
krono: ah, right
<cfbolz>
krono: that's why I said "to a first order of approximation" ;-)
<krono>
right
<cfbolz>
krono: anyway, I am still a bit skeptical that that kind of code is usefully shared via rlib
<cfbolz>
it's more a pattern
<cfbolz>
but the details are really different between languages
<krono>
right
<krono>
small_list differs betwen pycket, rsqueak and lamb/theseus, but only very very little
<cfbolz>
ok, cool
<krono>
the ugly thing is, I need 2**10 classes -.-
<cfbolz>
hehe
<krono>
ya, cause every other field could be an unboxed immutable int
<cfbolz>
the GC supports a meta field actually, so you can have different layouts
<cfbolz>
but the JIT doesn't
<krono>
you mean for rpython-level classes/
<cfbolz>
I think only for lltype structs
<cfbolz>
but yes
<krono>
the devil I know…
<cfbolz>
krono: in general it would be nice to have eg varsized instances at least. would save an indirection for eg tuples
<krono>
"variableSubclass" heh
<krono>
well, that's what small_list'ed classes do anyway, with make(...) but, again, that just a pattern
greedom has joined #pypy
<krono>
What was the connection between `_immutable_` and `_immutable_fields_` again?
<cfbolz>
krono: _immutable_ means "all fields, must not have any observable identity"
<krono>
ie, values?
<krono>
hum, neat thanks
<cfbolz>
it's not made use of too much
<krono>
seems so, but it is treated in small_list, so I better know what it's for :)
<cfbolz>
👍
lritter has quit [Quit: Leaving]
greedom has quit [Read error: Connection reset by peer]
greedom has joined #pypy
Julian has joined #pypy
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
greedom has quit [Remote host closed the connection]