<demon36[m]>
my implementation for the getString() is is doing too frequent native calls per non-single-byte-null string value retrieval
<demon36[m]>
I was thinking of adding a getZeroTerminatedByteArray() variant in jffi that takes null terminator width as an argument, for better performance, because its a bottleneck in my use case
<demon36[m]>
what do you guys think of that ?
joast has quit [*.net *.split]
joast has joined #jruby
<headius>
demon36 hello! Yes that seems reasonable. Would this go into the native library?
<demon36[m]>
headius assuming you mean jffi project, yes
<demon36[m]>
also do you think its better change the function getZeroTerminatedByteArray directly or create another override ?
<headius>
Probably another override
<headius>
This big issue though is getting new native builds. We have not modified the native code for many years because we do not have a way to rebuild it on several exotic platforms
<demon36[m]>
build process is very straightforward though, u might just leave that to the library user
<demon36[m]>
its good enough that you already provided prebuilt binaries for major platforms
<demon36[m]>
its good enough that you already provide prebuilt binaries for major platforms
<headius>
That is what we have done, perhaps we will be able to rely on this who have built in the past
<headius>
We have had many improvements on hold because of this
<demon36[m]>
good luck with that
<demon36[m]>
it appears now that I will have to override putZeroTerminatedByteArray too, in addition to checked variants
<headius>
Just so I fully understand...this needs native changes because it is too costly to look for the terminator before also reading the bytes