Atque has quit [Remote host closed the connection]
Atque has joined #pypy
mattip has quit [Ping timeout: 268 seconds]
mattip has joined #pypy
Atque has quit [Ping timeout: 268 seconds]
Atque has joined #pypy
lritter has quit [Quit: Leaving]
otisolsen70 has joined #pypy
jcea has joined #pypy
Dejan has joined #pypy
Dejan has quit [Client Quit]
Dejan has joined #pypy
j4at has joined #pypy
<j4at>
Why tuples doesn't have strategies like lists ?
<j4at>
other than 2 elements specialisation
<cfbolz>
j4at: because huge tuples with homogeneous types are rare
<cfbolz>
we try to motivate our optimizations with "in the wild" big programs that show problems
<Alex_Gaynor>
Many years ago I did a PoC for specializing arbitrary tuple shapes (size + types), but I don't think we ever merged it.
j4at has quit [Ping timeout: 252 seconds]
j4at has joined #pypy
<j4at>
There is only specilization for (int,int), (str, str) and (float, float), pretty sure that (str,int), (str, float), (int, str), (int, float) are very common too.
<j4at>
(int * 3) | (float * 3) | (int * 4) | (float * 4) are very common too for points in 3D/4D space or colors.
<exarkun>
What are the costs of having additional specializations?
j4at has quit [Read error: Connection reset by peer]
<Alex_Gaynor>
pessiminization of code that handles tuples polymorphically
<exarkun>
I understand this to mean there is an incremental runtime cost associated with each new specialization.
Atque has quit [Ping timeout: 268 seconds]
Atque has joined #pypy
dnn has joined #pypy
<dnn>
hello. is there anyone familiar with cffi here? the cffi doc says to ask here in addition to #cffi
jcea has quit [Ping timeout: 240 seconds]
<cfbolz>
exarkun: only if you use them in the same spot. but yes
catern has quit [Remote host closed the connection]