[PATCH 07 of 10] bytearray: replace array.array with bytearray

[timeless]

On Wed, May 18, 2016 at 12:08 PM, Martijn Pieters <mj at zopatista.com> wrote:
> Just pass in the integer result:
>   self._rbcrevs = bytearray(self._rbcrevslen * _rbcrecsize)
> This produces a bytearray of that size, filled with null bytes.

will do
> Would starting chunks as a bytearray, then extending *that* instead of
> a list, not work better? That would avoid creating a series of
> intermediate strings here.
>
>     chunks = bytearray()
>
>     for node in outgoing.missingheads:
>         # Don't compute missing, as this may slow down serving.
>         fnode = cache.getfnode(node, computemissing=False)
>         if fnode is not None:
>             chunks += node
>             chunks += fnode
>
> Now you have just one bytestring object to convert to str.

It definitely works. I'll include it with a v2.

>> -            deltatext = mdiff.textdiff(base, arraytext)
>> +            arraytext = bytearray(self.text())
>> +            deltatext = mdiff.textdiff(str(base), str(arraytext))
>>
>>          return arraytext, deltatext
>
> If self.text() is a string already, why convert to a bytearray then
> back to str?

I'll add a comment. The short answer is that mdiff.textdiff (~ bdiff)
currently takes a string,
but it shouldn't. It should take a bytearray.

The goal of this commit is to get rid of array.array. In a separate
commit, I'll convert bdiff to
accept bytearrays.

>> @@ -430,13 +427,12 @@
>>          try:
>>              data = repo.vfs.read(_fnodescachefile)
>>          except (OSError, IOError):
>>              data = ""
>> -        self._raw.fromstring(data)
>> +        data = repo.vfs.tryread(_fnodescachefile)
>> +        self._raw = bytearray(data)
>
> data is set twice now; you didn't remove the try..except.

Oops. That was a mis-merge. I didn't check the history when I did the merge.
I'll remove the first insertion.