From 17397ca0c2dad221dbcf50b41b76163c650c6eed Mon Sep 17 00:00:00 2001 From: James Bowes Date: Wed, 22 Aug 2012 11:37:32 -0300 Subject: [PATCH] Add start of algorithm file --- ALGORITHM.md | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 114 insertions(+) create mode 100644 ALGORITHM.md diff --git a/ALGORITHM.md b/ALGORITHM.md new file mode 100644 index 0000000..f665b5e --- /dev/null +++ b/ALGORITHM.md @@ -0,0 +1,114 @@ +PACKING ALGORITHM +================= + +Begin with a list of content set paths: +''' +/content/dist/rhel/$releasever/$basearch/os +/content/dist/rhel/$releasever/$basearch/debug +/content/dist/rhel/$releasever/$basearch/source/SRPMS +/content/dist/jboss/source +/content/beta/rhel/$releasever/$basearch/os +/content/beta/rhel/$releasever/$basearch/debug +/content/beta/rhel/$releasever/$basearch/source/SRPMS +''' + +Create a prefix tree out of the paths, where each node in the tree contains the +names of its children, rather than the children containing its own name: + +''' ++-------+ +------+ +| | | | +|-------+ |------| +|content+--+-------+ |source+-+---+ ++-------+ | | +------+ | | + |-------| | |---| + |dist +--------+-------+ | +---+ + +--+beta | | | | + | +-------+ |-------| | + | |jboss +--+ ++-------+ |rhel +--+ +| | +-------+ | +|-------| | +|rhel +--+-----------+ +-----------+ ++-------+ | | | | + |-----------| |-----------| ++----------+$releasever| |$releasever+--+ +| +-----------+ +-----------+ | +| | ++---------+ +---------+ +| | | | +|---------| |---------| +|$basearch+--+ +--+$basearch| +---+ ++---------+ | | +---------+ | | + | | |---| + +--------+ +-------+ +---+ + | | | | | + |--------| |-------| | ++---+--------+os | |os +---------+ +| | +--+debug | |debug +-----------+---+ +|---| | |source +--+ |source +--+ | | ++---+ | +--------+ | +-------+ | |---| + | | | +---+ + +---+ +------+ | + | | | | +-----+ + |---| |------| | | + +---+ +--+SRPMS | |-----| + | +------+ |SRPMS+--+---+ + | +-----+ | | + +---+ |---| + | | +---+ + |---| + +---+ +''' + +This eliminates the duplication in the prefixes of our paths. We now only have +one instance of the word "content", for example. We now find any duplicate +subtrees, remove the duplicate, and point all references from the duplicate to +the original: + +''' ++-------+ +------+ +| | | | +|-------+ |------| +|content+--+-------+ |source+-+ ++-------+ | | +------+ | + |-------| | | + |dist +--------+-------+ | | + +--+beta | | | | | + | +-------+ |-------| | | + | |jboss +--+ | ++-------+-------------------+rhel | | +| | +-------+ | +|-------| | +|rhel +--+-----------+ | ++-------+ | | | + |-----------| | ++----------+$releasever| | +| +-----------+ | +| | ++---------+ | +| | | +|---------| | +|$basearch+--+ | ++---------+ | | + | | + +--------+ | + | | | + |--------| | + +-------+os | | + | +--+debug | | + | | |source +--+ | + | | +--------+ | | + | | | | + | | +------+ | + | +-------+ | | | + | | |------| | + | | +--+SRPMS | | + | | | +------+ | + | | | | + +------------+---+-------------------------+ + | | + |---| + +---+ +''' +