Hi, after read through the source code, I'm a little confused about why we need to mmap the current datafile to memory, even though we never use the mmap reader when the datafile is not readonly.
Then only usage of the mmap reader is at the datafile.ReadAt()
But we create a mmap Reader everytime here in the Newdatafile()
ra,err=mmap.Open(fn)iferr!=nil{returnnil,err}
Hi, after read through the source code, I'm a little confused about why we need to mmap the current datafile to memory, even though we never use the mmap reader when the datafile is not readonly.
Then only usage of the mmap reader is at the datafile.ReadAt()
```golang
if df.w == nil {
n, err = df.ra.ReadAt(b, index)
} else {
n, err = df.r.ReadAt(b, index)
}
```
But we create a mmap Reader everytime here in the Newdatafile()
```golang
ra, err = mmap.Open(fn)
if err != nil {
return nil, err
}
```
Yeah, I figured. It just seems that the mmap reader is never used when the datafile is the current datafile, which is not readonly. Since the mmap package using the mmap with PROT_READ option, the mmap reader is meant for readonly purpose. However, for the current datafile, the read/write option is always through the fd got from the Open syscall, so I think it might be better not set the ra for the current datafile.
Yeah, I figured. It just seems that the mmap reader is never used when the datafile is the current datafile, which is not readonly. Since the mmap package using the mmap with PROT_READ option, the mmap reader is meant for readonly purpose. However, for the current datafile, the read/write option is always through the fd got from the Open syscall, so I think it might be better not set the ra for the current datafile.
Hi, after read through the source code, I'm a little confused about why we need to mmap the current datafile to memory, even though we never use the mmap reader when the datafile is not readonly.
Then only usage of the mmap reader is at the datafile.ReadAt()
But we create a mmap Reader everytime here in the Newdatafile()
Hmmm thanks for reporting this, I'll have a close look! The intent behind memory mapping the datafiles was for performance
Yeah, I figured. It just seems that the mmap reader is never used when the datafile is the current datafile, which is not readonly. Since the mmap package using the mmap with PROT_READ option, the mmap reader is meant for readonly purpose. However, for the current datafile, the read/write option is always through the fd got from the Open syscall, so I think it might be better not set the ra for the current datafile.
Yeah I think you're right actually. Did you want to put a PR up for this since you discovered it? 🤔 You should get the credit for fixing this 🤗
Thanks, I'll give it a try.