use bytes.Buffer copy instead of make slice #2

[u5surf]

#2
I fix to be using io.Copy from reader stream to byte.Buffer instead of making slice object.
And I would like to indicate memory usage performance. Let me how to measure the performance which you tested.

[alldroll]

For measure the performance you can use built in go prof tools or something like this

/usr/bin/time -lp (or -v)  ./dump data.cdb > /dev/null

data.cdb you can create by yourself with any csv file (with 2 columns) or use already created cdb file for instance https://github.com/alldroll/suggest/blob/master/testdata/db/cars.cdb

[alldroll]

I don't think that this approach could help to optimize memory usage. Every loop step you allocate a bytes.Buffer object and io.Copy also allocates []byte. I think such code as

  var (
         key   = []byte{} // better to allocate with some capacity
         value = []byte{}
  )
  ....
  loop step
  ....

  if len(key) < int(keySize) {
       key = make([]byte, keySize*2)
  } // otherwise we save memory for key reading

  // the same for value reading
  ...
  
  err = csvWriter.Write([]string{string(key[:keySize]), string(value[:valSize])})

could help in the issue.

[alldroll]

It's a bad practice to ignore errors