From: Brian Warner Date: Wed, 28 Mar 2007 19:52:44 +0000 (-0700) Subject: update memory-footprint and share-size docs in ICodecEncoder X-Git-Url: https://git.rkrishnan.org/%5B/frontends/flags/$rel_link?a=commitdiff_plain;h=74d08399804640a6200592b8612ac48c2aa0944e;p=tahoe-lafs%2Ftahoe-lafs.git update memory-footprint and share-size docs in ICodecEncoder --- diff --git a/src/allmydata/interfaces.py b/src/allmydata/interfaces.py index fc7965f7..30d1257f 100644 --- a/src/allmydata/interfaces.py +++ b/src/allmydata/interfaces.py @@ -217,17 +217,24 @@ class ICodecEncoder(Interface): For each call, encode() will return a Deferred that fires with two lists, one containing shares and the other containing the shareids. - The get_share_size() method can be used to determine the length of the - share strings returned by encode(). - + The get_share_size() method can be used to determine the length of + the share strings returned by encode(). Each shareid is a small + integer, exactly as passed into 'desired_share_ids' (or + range(max_shares), if desired_share_ids was not provided). + The shares and their corresponding shareids are required to be kept together during storage and retrieval. Specifically, the share data is useless by itself: the decoder needs to be told which share is which by providing it with both the shareid and the actual share data. - The memory usage of this function is expected to be on the order of + This function will allocate an amount of memory roughly equal to:: + + (max_shares - required_shares) * get_share_size() - (max_shares - required_shares) * get_share_size(). + When combined with the memory that the caller must allocate to + provide the input data, this leads to a memory footprint roughly + equal to the size of the resulting encoded shares (i.e. the expansion + factor times the size of the input segment). """ # rejected ideas: