corres wrote: ↑Tue Dec 03, 2019 12:02 pm
I use VS 2015 and auxiliaries for compilation.
Maybe the current src is not portable opposite to the earlier.
My compiler shows faults in version.h, \neural\encoder.cc, \utils\esc_codes.cc (undeclared identifiers, fault in number of arguments, unresolved externals, etc. )
Could you provide more data? Like, compiler error messages?
In any case, I suggest joining the discord. It would be much easier to investigate why you have problems building the code this way.
Sorry, I can not join discord.
I uploaded "log.zip" file on http://wikisend.com
File ID 640040
Password leela
This was the first stop of compilation.
Some note
I make compilation offline. This did not cause any issues before. Seeing the name of lc0 on chess GUI I am afraid you may connect your compiler to Github.com. I do not know why you did it, what is the benefit doing this, but I do not want to endanger my machine to any viruses.
corres wrote: ↑Tue Dec 03, 2019 12:02 pm
I use VS 2015 and auxiliaries for compilation.
Maybe the current src is not portable opposite to the earlier.
My compiler shows faults in version.h, \neural\encoder.cc, \utils\esc_codes.cc (undeclared identifiers, fault in number of arguments, unresolved externals, etc. )
Could you provide more data? Like, compiler error messages?
In any case, I suggest joining the discord. It would be much easier to investigate why you have problems building the code this way.
Sorry, I can not join discord.
I uploaded "log.zip" file on http://wikisend.com
File ID 640040
Password leela
This was the first stop of compilation.
Some note
I make compilation offline. This did not cause any issues before. Seeing the name of lc0 on chess GUI I am afraid you may connect your compiler to Github.com. I do not know why you did it, what is the benefit doing this, but I do not want to endanger my machine to any viruses.
Did you regenerate your .sln file using build-cuda.cmd?
There are files added and removed between releases, and some of build options change too, so it's not possible to just copy new files over and use the same generated build config.
You can fetch project into a fresh directory and run build-cuda.cmd from there.
During the build configuration step, the script may indeed need to download some dependencies from the internet, e.g. zlib or protobuf.
crem wrote: ↑Tue Dec 03, 2019 4:54 pm
...
Did you regenerate your .sln file using build-cuda.cmd?
There are files added and removed between releases, and some of build options change too, so it's not possible to just copy new files over and use the same generated build config.
You can fetch project into a fresh directory and run build-cuda.cmd from there.
During the build configuration step, the script may indeed need to download some dependencies from the internet, e.g. zlib or protobuf.
Between two run of compilations I always delete the full \build\ library, together with lc0.sln.
I downoaded zlib and protobuf-s manually.
Where from you get the "BUILD IDENTIFIER" what lacks from source? Maybe from Github?
In the ..\neural\encoder.cc there is a "switcher" function in which the second branch (case....::INPUT_112_WITH_CASTLING_PLANE:....) has no any connection and it is defective.
What is the task of it? (maybe nothing)
corres wrote: ↑Tue Dec 03, 2019 6:30 pm
...
Where from you get the "BUILD IDENTIFIER" what lacks from source? Maybe from Github?
In the ..\neural\encoder.cc there is a "switcher" function in which the second branch (case....::INPUT_112_WITH_CASTLING_PLANE:....) has no any connection and it is defective.
What is the task of it? (maybe nothing)
Meanwhile I found the answers.
"BUILD IDENTIFIER" is generated by meson. I no need such a thing so I do not repair meson.build file and I use the files from version 0.22 for version.h.
Sorry, but I did not know you use newer .proto files for lczero-common. After I checked them I downloaded and copied them to \libs\.
So remained the issue about colored marks but I am not interested in them.