Closed noloader closed 6 years ago
The error simply indicates that ${CMAKE_SYSTEM_NAME}
is blank. That variable is filled with the result of uname -s
by cmake.
EDIT:
I think I have an idea why this is happening... that variable is only populated after the project(...)
cmake function is called, and in the following lines, that function is not called if cmake version is below 3.0.0:
if (${CMAKE_VERSION} VERSION_LESS "3.0.0")
set(cryptopp_VERSION_MAJOR 7)
set(cryptopp_VERSION_MINOR 0)
set(cryptopp_VERSION_PATCH 0)
else ()
cmake_policy(SET CMP0048 NEW)
project(cryptopp VERSION 7.0.0)
if (NOT ${CMAKE_VERSION} VERSION_LESS "3.1.0")
cmake_policy(SET CMP0054 NEW)
endif ()
endif ()
Edit 2: Just pushed commit 355cee063d94aaa85ab82a8b3334552c5a79d339 to fix that.
Edit 2: Just pushed commit 355cee0 to fix that.
@noloader can you please verify this has been fixed and close the issue if so?
Solaris build fixed as of commit ed5c3deda48a101fd26c9a911d8429434775d7c7
Good job @abdes,
You just poked Oracle in the eye. Distributing old components is part of Oracle's business model. Oracle charges a lot of money to access their updated software repo. Many folks have pay the extortion, but not because of Crypto++...
Here's the SunCC (Sun C++ compiler) test case. It works fine, too:
cd cryptopp
cp TestScripts/cryptest-cmake.sh .
CXX=/opt/developerstudio12.6/bin/CC ./cryptest-cmake.sh
You can do the same directly from the command line with:
cd cmake_build
CXX=/opt/developerstudio12.6/bin/CC cmake ../
...
You just poked Oracle in the eye. Distributing old components is part of Oracle's business model. Oracle charges a lot of money to access their updated software repo. Many folks have pay the extortion, but not because of Crypto++...
I've always had an issue with Sun since school days when they absolutely refused to adopt Motif as a windowing system. It has not changed for 25 years, it even became worse with Oracle. Solaris X86 is ugly, slow, and archaic, by laziness or to just lock the customers into bad stuff. BUt, like you said, we gotta do we gotta do and make the stuff people need available even for those unfortunate ones who are still locked into archaic platforms :-)
I'm testing on GCC112 on the compile farm:
Here is the line in CmakeList.txt:
ping @abdes. Would you mind taking a look at this? It looks like the break was introduced before we created this repository. The pain point here is, (1) CMake does not define the variable that is being used and (2) there's no way I know of to get a portable terminal to run commands (assuming you want to run
uname
).Also see GCC Compile Farm Accounts for Testing.