1 - keep history of installed packages/journal of package transaction,
2 so we can roll back to yesterday, or see what got installed in the
7 - gzip repository of look-aside pkg xml files somehow?
9 - transactions, proper recovery, make sure we don't poop our package
10 database (no more rm /var/lib/rpm/__cache*).
12 - diff from one package set to another answers: "what changed in
13 rawhide between since yesterday?"
15 - rewrite qsort and bsearch that doesn't require global context var
16 and can output a map describing the permutaion.
18 - use hash table for package and property lists so we only store
19 unique lists (like for string pool).
21 - use existing, running system as repo; eg
23 razor update razor://other-box.local evince
25 to pull eg the latest evince and dependencies from another box. We
26 should be able to regenerate a rzr pkg from the system so we can
27 reuse the signature from the originating repo.
29 - Ok, maybe the fastest package set merge method in the end is to use
30 the razor_importer, but use a hash table for the properties. This
31 way we can assign them unique IDs immediately (like tokenizing
34 - test suite should be easy, just keep .repo files around and test
35 different type of upgrades that way (obsoletes, conflicts, file
36 conflicts, file/dir problems etc). Or maybe just keep a simple file
37 format ad use a custom importer to create the .repo files.
39 - pipelined download and install; download is network bound, install
40 is disk bound. Start installing once we have self-contained set of
41 packages. Install in reverse topo-sort order. Interruptible
42 installation; stops at nearest checkpoint.
44 - make packages pointers be either an index into the package pool or a
45 direct link to a package when there is only one package. set a high
46 bit to indicate which it is. similar for properties.
48 - split out hash table code from importer, make the merger use just
53 do { ... } while (((e++)->name & RAZOR_ENTRY_LAST) == 0);
55 idiom for iteration of directories.