This accomplishes quite a few things with one rather invasive change.
1. Iteration is much more performant, due to a reduction in pointer
chasing and linear item access.
2. Data structures are smaller- we no longer have the overhead of the
linked list as the file struts are now laid out consecutively in
memory.
3. Memory allocation has been massively reworked. Before, we would
allocate three different pieces of memory per file item- the list
struct, the file struct, and the copied filename. What this resulted
in was massive fragmentation of memory when loading filelists since
the memory allocator had to leave holes all over the place. The new
situation here now removes the need for any list item allocation;
allocates the file structs in contiguous memory (and reallocs as
necessary), leaving only the strings as individually allocated. Tests
using valgrind (massif) show some pretty significant memory
reductions on the worst case `pacman -Ql > /dev/null` (366387 files
on my machine):
Before:
Peak heap: 54,416,024 B
Useful heap: 36,840,692 B
Extra heap: 17,575,332 B
After:
Peak heap: 38,004,352 B
Useful heap: 28,101,347 B
Extra heap: 9,903,005 B
Several small helper methods have been introduced, including a list to
array conversion helper as well as a filelist merge sort that works
directly on arrays.
Signed-off-by: Dan McGee <dan@archlinux.org>
This allows us to capture size and mode data when building filelists
from package files. Future patches will take advantage of this newly
available information, and frontends can use it as well.
Signed-off-by: Dan McGee <dan@archlinux.org>
This removes the need to write accessor methods for every type we have,
and simplifies the API. Any type that doesn't need magic* can be
converted in this fashion to make it easier for frontend applications to
use, as well as make it less of a pain to introduce new such structs in
the future.
* "magic" meaning something like pmpkg_t where values can be lazy loaded.
Signed-off-by: Dan McGee <dan@archlinux.org>
This kills a lot more global handle business off. sync.c still requires
the handle declaration for one reference that can't be changed yet; it
will be removed in a future patch which isolates all of the necesary API
changes.
Signed-off-by: Dan McGee <dan@archlinux.org>
* Make conflict_isin() static; it is used nowhere else.
* Remove does_conflict(): it turns out to be replaceable by a single call to
_alpm_depcmp(). By pushing it up, we can reduce calls to _alpm_splitdep()
from 60,368 to 16,940 during one test -Su operation I ran.
Signed-off-by: Dan McGee <dan@archlinux.org>
Sometimes "foo conflicts with bar" information is not enough, see this
thread: http://bbs.archlinux.org/viewtopic.php?id=77647. That's why I added
a new reason field to our pmconflict_t struct that stores the packager-
defined conflict that induced the fact that package1 conflicts with
package2.
I modified the front-end (in callback.c, sync.c, upgrade.c) to print this
new information as well.
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
From now on _alpm_db_find_fileconflicts() works with upgrade and remove
target lists (like checkdeps), which makes it transaction independent
(we still need a trans param because of the progressbar). This is a small
step towards the universal transaction. So we call this function directly
from sync.c before commiting the remove transaction. This is much safer,
but we can get false fileconflict error alarms in some tricky cases
("symlinks puzzle" etc).
The patch on find_fileconflict looks complex, but it is mainly an
"indent-patch", the new code-part can be found after the
/* check remove list ... */ comment, and I modified something around the
"file has changed hand" case (see comment modifications in the code).
Unfortunately sync.c became more ugly, because we have to create 2 parallel
internal transactions: to avoid duplicated work, upgrade transaction is
used to load package data (filelists). This problem will disappear, when
we finally get rid of internal transactions.
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Dan McGee <dan@archlinux.org>
This function has a limited purpose, but might be interesting to do a
sanity check from a frontend (eg testdb).
Also removed the private _alpm_checkconflicts function to avoid confusion.
This function was used only once in libalpm, in sync.c, and was just a
single line anyway. Having to do it manually makes it explicit that we are
looking for two kind of conflicts (targ vs targ and db vs targ).
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
This also affects all structures with static strings, such as depmiss,
conflict, etc. This should help a lot with memory usage, and hopefully make
things a bit more "idiot proof".
Currently our pactest pass/fail rate is identical before and after this
patch. This is not to say it is a perfect patch- I have yet to pull valgrind
out. However, this should be quite safe to use in all situations from here
on out, and we can start plugging the memleaks.
Original-work-by: Aaron Griffin <aaronmgriffin@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
Update the GPL boilerplate to direct people to the GNU website for a copy of
the license, as well as bump all of Judd's copyrights to 2007.
Signed-off-by: Dan McGee <dan@archlinux.org>
_alpm_innerconflicts: check for target<->target conflicts
_alpm_outerconflicts: check for target<->localpkg conflicts
This will be useful in sync.c clean-up and in testdb.c
As an application the patch also fixes a misleading message (and a memleak)
in add.c
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
pmdepmissing_t was used for two totally different things :
missing dependencies, and dependency conflicts.
So this patch simply adds a type for dep conflicts,
and convert the code to use it.
This fix the TODO in conflict.c :
/* TODO WTF is a 'depmissing' doing indicating a conflict? */
Additionally, the code in conflict.c now eliminates the duplicated conflicts.
If pkg1 conflicts with pkg2, and pkg2 conflicts with pkg1, only one of them will be stored.
However the conflict handling in sync_prepare (sync.c) is still very asymetrical, and very ugly too.
This should be improved in the future (there is already a pending patch from Nagy that cleans it a lot).
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
The names related to conflicts are misleading :
For dependencies conflicts, the type is pmdepmissing,
and the function names contain just "conflict".
For file conflicts, the type is pmconflict,
and some functions contained just "conflict", some others "fileconflict".
So this is the first step for improving the situation.
Original idea/patch from Nagy, but the patch already didn't apply anymore,
so I did it again.
The main difference is that I kept the conflictype, with the following renaming :
pmconflicttype_t -> pmfileconflicttype_t
PM_CONFLICT_TYPE_TARGET -> PM_FILECONFLICT_TARGET
PM_CONFLICT_TYPE_FILE -> PM_FILECONFLICT_FILESYSTEM
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
reordering and adding ones that were forgotten (noticed when trying to
compile after reordering).
* Updated the HACKING file to include information on #include usage.
* print -> vprint in "making dir" function in pactest.
from one package to another seemlessly (knock on wood). This is implemented
through the use of two skip lists in the trans struct- skip_add and
skip_remove, which replace the former trans->skiplist.
* Removed an unnecessary function parameter, added a necessary one.
* If a package has no backup files, print '(none)' under the heading so it is
more obvious.
* Updated my TODO list.
optimizations in order to make the code much more readable and
type-checkable. Every enum in the library now has it's own type that
should be used instead of the generic 'unsigned char'. In addition,
several #define statements dealing with constants were converted to
enums.
Signed-off-by: Dan McGee <dpmcgee@gmail.com>
* renamed pmlist_t -> alpm_list_t
* made alpm_list_t a public type (alpm_list.h header)
* removed additional storage for registered DBs in pacman source
* some code cleanup
* removed duplicate (pm)list_display functions from pacman source
* misc code cleanup
* libalpm api changes - move from a _getinfo(p, WHAT_WE_WANT) scheme to a
typesafe _get_what_we_want(p) scheme [not 100% complete yet]
* some const correctness changes
* removal of PM_* types in alpm.h in favor of the pm*_t types used throughout
libalpm
* Removed the PMList typedef, in favor of the same naming scheme other
structs use 'pmlist_t'
* Added a time stamp on debug output, to make it more informational
* Moved alpm_db_register to _alpm_db_register, making the public function
not take a callback parameter
