You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
compiling mime-roundtrip exhausts virtual memory on mips #381
This is virtual memory address space, so we can't just "throw more memory" at it. So this will affect the real builds as well. Unfortunately Debian does not have 64-bit mips userland at this time.
Do you see any obvious/easy ways for reducing the memory usage of that build command?
This was reported as https://bugs.debian.org/754258 and fixed by adding --param ggc-min-expand=20 as compiler flags. In theory cpp-netlib could add this to the CMake build too.
+1 to @mclow -- @infinity0 do you mind sending a pull request to 0.11-devel in time for the next release (which should happen at the end of this week)?
That's fine, thanks @infinity0 -- it looks like there's a few more things that need to be fixed by me anyway for 0.11.1RC1 which should happen by end of week, but I have no illusions that that would be the final RC. :)
Another option is to unconditionally add set (CMAKE_CXX_FLAGS "--param ggc-min-expand=20") to mime's CMakeLists.txt. It works on my computer but I've not tested it with others. Up to you guys.
On second thoughts, probably better to use set_source_files_properties( mime-roundtrip.cpp PROPERTIES COMPILE_FLAGS "--param ggc-min-expand=20" ) so it's just for that one file.
Activity
infinity0 commentedon Feb 9, 2014
Pinging @mclow as I understand you're the author of this part.
infinity0 commentedon Aug 19, 2014
This was reported as https://bugs.debian.org/754258 and fixed by adding
--param ggc-min-expand=20as compiler flags. In theory cpp-netlib could add this to the CMake build too.mclow commentedon Aug 19, 2014
That sounds reasonable to me
deanberris commentedon Aug 20, 2014
+1 to @mclow -- @infinity0 do you mind sending a pull request to 0.11-devel in time for the next release (which should happen at the end of this week)?
deanberris commentedon Sep 1, 2014
Ping @infinity0 -- Do you think this is OK to close, or will you try and send something that could make it into 0.11-devel and thus part of 0.11.1?
infinity0 commentedon Sep 1, 2014
i don't think i will get around to it myself this week, but could leave the issue open?
deanberris commentedon Sep 1, 2014
That's fine, thanks @infinity0 -- it looks like there's a few more things that need to be fixed by me anyway for 0.11.1RC1 which should happen by end of week, but I have no illusions that that would be the final RC. :)
infinity0 commentedon Sep 14, 2014
This is apparently non-trivial to do in CMake so I'll skip it for now.
http://stackoverflow.com/questions/11944060/how-to-detect-target-architecture-using-cmake
infinity0 commentedon Sep 14, 2014
Another option is to unconditionally add
set (CMAKE_CXX_FLAGS "--param ggc-min-expand=20")to mime'sCMakeLists.txt. It works on my computer but I've not tested it with others. Up to you guys.infinity0 commentedon Sep 14, 2014
On second thoughts, probably better to use
set_source_files_properties( mime-roundtrip.cpp PROPERTIES COMPILE_FLAGS "--param ggc-min-expand=20" )so it's just for that one file.deanberris commentedon Sep 15, 2014
I'm happy with just setting the properties on the single file if that's that only one causing the problem.