1 Towards replacing rpm + yum (0.1):
5 - pre install check; check that dirs can be created (no files where
6 want to create dirs), move config files according to file
9 - store rpm headers for installed packages.
11 - rpm seems to consider glibc > 2.6.90 to mean greater than
12 2.6.90-anything. That is, a comparison that doesn't mention the
13 release field, shouldn't regard the release field of pkgs it
14 compares against. glibc-common-2.6.90 has
16 conflicts: glibc < 2.6.90, glibc > 2.6.90
18 since rpm doesn't let you do glibc != 2.6.90, and
20 requires: glibc = 2.6.90
22 will always pull in glibc. But even with a != relation, would
23 glibc-2.6.90-16 be equal to 2.6.90? glibc 2.7.90-8 dropped it in
24 favor of requires = 2.7.90-8 (#225806).
28 - figure out how to canonically represent empty string... ~0?
30 - space calculation before transaction, but ideally, do a number of
33 - pre-link changing binaries and libs on disk screwing up checksum?
35 - nail down byte-order of repo file.
37 - version the sections in the file, put the element size in the header
38 so we can add stuff to elements in a backwards compatible way.
39 maybe not necessary, we can just add sections that augment the
40 sections we want to add to (similar to how rpm has add versioned
43 - pipelined download and install; topo-sort packages in update set,
44 pick one with all deps in the current set, add it to the current set
45 and satisfy deps against update set => result: minimal update
46 transaction. Queue download and install/update transaction for the
47 packages in the minimal set, start over. This also makes the
48 installation phase much more interruptible, basically just stop
49 after a sub-transaction finishes. As we keep the update set around
50 as a target, we can restart if needed. Probably don't need to, can
51 just do a new update. During a sub-transaction we should keep the
52 target set (i.e. the current set to be) around as a lock file
53 (system.repo.lock or so, see git) so that razor updates are
54 prevented if the systems crashes during an update.
58 - keep history of installed packages/journal of package transaction,
59 so we can roll back to yesterday, or see what got installed in the
62 - transactions, proper recovery, make sure we don't poop our package
63 database (no more rm /var/lib/rpm/__cache*).
65 - use hash table for package and property lists so we only store
66 unique lists (like for string pool).
68 - use existing, running system as repo; eg
70 razor update razor://other-box.local evince
72 to pull eg the latest evince and dependencies from another box. We
73 should be able to regenerate a rzr pkg from the system so we can
74 reuse the signature from the originating repo.
76 - Ok, maybe the fastest package set merge method in the end is to use
77 the razor_importer, but use a hash table for the properties. This
78 way we can assign them unique IDs immediately (like tokenizing
81 - test suite should be easy, just keep .repo files around and test
82 different type of upgrades that way (obsoletes, conflicts, file
83 conflicts, file/dir problems etc). Or maybe just keep a simple file
84 format ad use a custom importer to create the .repo files.
88 do { ... } while (((e++)->name & RAZOR_ENTRY_LAST) == 0);
90 idiom for iteration of directories.
92 - overlay package sets? mount a read-only /usr over nfs or from the
93 virt-host and have a local package set overlaid over the read-only
94 one. shouldn't need new features in the core package set data
95 structure, but should be just conventions on top. we have the base
96 package set from the r/o system, the overlay set from the local
97 system and we can have an effective package set which is the merge
98 of everything from the overlay into the base set. the effective set
99 is easy to compute and we could do it on the fly or cache it. or
100 maybe the effective set is the on-disk representation and the
101 overlay can be computed when needed, we just keep a link back to the
104 - incremental rawhide repo updates? instead of downloading 10MB zipped
105 repo every time, download a diff repo? Should be pretty small,
106 especially if we don't have file checksums in metadata. Filenames
107 and properties are for the most part already present, typically just
108 a version bump plus maybe tweaking a couple requires. The upstream
109 repo can store multiple incremental updates in one big file and
110 provide an index file that maps updates for a given date (we should
111 use repo-file checksums though) to a range in the file: Download the
112 index file, search for a match for your latest rawhide.repo file,
113 download range of updates that brings it up to date.
115 - use hash tables for dirs when importing files to avoid qsorting all
120 - eliminate duplicate entries in package property lists.