[win32] Permission denied in fancy_rename

[John Arbash Meinel]

Aaron Bentley wrote:
> John Arbash Meinel wrote:
>> Anyway, I think get_bytes/put_bytes is good, but we should think about
>> the memory impact.
> 
> Perhaps get_bytes/put_bytes should take/produce an iterable?  That would
> ensure a low memory impact.
> 
> Strings, files, and string iterators are all iterables of strings, and
> we can special-case files and strings if we want.
> 
> Aaron
> 

You've been trying to get iterators for a while now, and this might just
be the place for it.

We certainly could have the default 'get_bytes' something like:

def get_bytes(relpath):
  fp = self.get(relpath)
  chunk = fp.read(32768)
  while chunk:
    yield chunk
    chunk = fp.read(32768)

And if we have some sort of a string, we can either do:

def get_bytes(relpath):
  s = something
  return [s]

Or some sort of split based on length.

http could just yield chunks from the urllib.urlopen() file. (It seems
to be a wrapper named addinfourl (add info url) around a
socket.makefile() call).

sftp could likewise return chunks from the SFTP file.


Then the naive put_bytes could become:

def put_bytes(relpath, bytes, mode=None):
  buffer = StringIO()
  for b in bytes:
    buffer.write(b)
  self.put(relpath, buffer, mode=mode)


I could see special casing "put_bytes" for strings so we don't iterate
over every character. Simply wrapping it in a list is probably enough.

John
=:->


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060503/831fb3d9/attachment.pgp