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
Can authors write guides to introduce how we can use this project in the Windows environment? #33
Unfortunately, the project only supports Linux right now, and I've updated the README.md to reflect this.
However, I believe the only incompatibilities with Windows are NanoLog's usage of POSIX AIO/threads and the assembly rdstc instruction (which can be replaced with std::high_resolution_clock). If you like to make a port and submit it as a pull-request, I would be happy to accept it.
font{
line-height: 1.6;
}
ul,ol{
padding-left: 20px;
list-style-position: inside;
}
Hi Dear Mr Yang,
NanoLog is a greate project, thank you for your reply!Best Wishes,Erwin Smith
On 7/2/2020 05:57，syang0<notifications@github.com> wrote：
Hi thingsareright,
Unfortunately, the project only supports Linux right now, and I've updated the README.md to reflect this.
However, I believe the only incompatibilities with Windows are NanoLog's usage of POSIX AIO/threads and the assembly rdstc instruction (which can be replaced with std::high_resolution_clock). If you like to make a port and submit it as a pull-request, I would be happy to accept it.
Best Regards,
Stephen Yang
—You are receiving this because you authored the thread.Reply to this email directly, view it on GitHub, or unsubscribe.
What were the reasons for using posix AIO? Asynchrony is emulated. This doesn't look optimal (just open the glibc code to make sure). In musl in General each operation will create its own thread
The decision was made long ago, so I may not remember all the details. However, I believe posix AIO was originally chosen because it was more convenient at the time.
The only other option I had considered back then was the kernel's libaio, but that had requirements about memory alignment and (I think) performed a memory copy in the calling thread. For posix AIO, all I had to do was give it the pointer and byte size of a buffer, and the library would handle the rest; so I chose posix AIO.
Activity
syang0 commentedon Jul 1, 2020
Hi thingsareright,
Unfortunately, the project only supports Linux right now, and I've updated the README.md to reflect this.
However, I believe the only incompatibilities with Windows are NanoLog's usage of POSIX AIO/threads and the assembly rdstc instruction (which can be replaced with std::high_resolution_clock). If you like to make a port and submit it as a pull-request, I would be happy to accept it.
Best Regards,
Stephen Yang
thingsareright commentedon Jul 2, 2020
Roman-Koshelev commentedon Nov 15, 2020
What were the reasons for using posix AIO? Asynchrony is emulated. This doesn't look optimal (just open the glibc code to make sure). In musl in General each operation will create its own thread
syang0 commentedon Nov 21, 2020
Hi Roman-Koshelev,
The decision was made long ago, so I may not remember all the details. However, I believe posix AIO was originally chosen because it was more convenient at the time.
The only other option I had considered back then was the kernel's libaio, but that had requirements about memory alignment and (I think) performed a memory copy in the calling thread. For posix AIO, all I had to do was give it the pointer and byte size of a buffer, and the library would handle the rest; so I chose posix AIO.
Best Regards,
Stephen Yang