| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
|
|
references to the apt cache.
This doesn't totally fix the problem with the program crashing if it's
exited while changelogs are downloading, but it makes it a lot easier
to fix. The reason is that I can now leave jobs in the queues while
the cache is closed, and resume processing when it's reopened: when
cache structures were potentially hanging around, that wasn't an option.
|
|
separate file, so it can be reused elsewhere.
|
|
parsing when displaying the upgrade changelog list.
It turns out that the reason aptitude took so long to display the
changelogs in my upgrade set is that extracting changelogs from the
cache and parsing them takes a long time. Now aptitude can also cache
the parsed version of the changelog, bypassing those steps entirely,
although currently only the GTK+ interface does so. (the other
interfaces don't really need this at the moment anyway)
|
|
The code it invokes has to run in the main thread for safety reasons.
Without this we get random crashes.
|
|
The setup phase is somewhat I/O intensive; this avoids blocking the UI
while it's going on. Also, this converts the ui_download_manager and
ui_download_thread classes to using boost::shared_ptr to store the
download manager; that should make its memory management a lot simpler.
|
|
installed.
|
|
number of threads and parse jobs when there are lots of available upgrades.
|
|
(Closes: #549897)
|
|
|
|
using libparse-debianchangelog-perl.
As noted by Sven Joachim, dpkg-parsechangelog is in the "dpkg-dev"
package, not a package users usually have installed. That package
is quite a bit larger than libparse-debianchangelog-perl, and
dpkg-parsechangelog runs about four times slower than parsechangelog,
so I think it's better to use the "nonstandard" changelog parser here.
|
|
information object, create an explicitly named value and hide the default constructor.
Thanks to Miroslav Kure for the initial patch.
|
|
translate the empty string.
|
|
Kure for pointing this out).
|
|
Over a year and a half ago, the options I needed from
libparse-debianchangelog-perl were finally added to dpkg-parsechangelog.
Being able to parse changelogs without asking the user to install
optional software is great!
|
|
c8aec0d44f84
|
|
comparable across runs.
Not every set of choices is sorted yet; there might be a few more of
these to write.
|
|
than iterators.
This cuts the runtime memory use down by 20-30%. The problem was that
a *huge* number of these objects are allocated, and iterators store
members that either aren't needed, or are redundant with other members
of the containing resolver object.
The dep-cache pointers could, perhaps, be dropped, if I forced
the routines that want one to request a "universe" pointer instead.
On the one hand, that's just about everything now; on the other hand,
that would chop many of these structures down to a single pointer!
Worth a thought, anyway.
|
|
I think this is a bit clearer and more descriptive.
|
|
|
|
It's not really an explanation of the solution but rather of an individual action. Bonus: it's shorter.
|
|
The examined solution is an element of the list of existing solution, so should be on the right of it.
Solution view mode now horizontal to save a little vertical space.
The 'next solution' and 'apply' are now right below the examined solution, much more logical and save mouse movements when moving between choices and 'next'.
|
|
rebuilding package lists after a search.
This actually always uses the entry as a progress bar; I decided that
was probably less confusing than sometimes using the entry, and using
the global progress bar when the list was rebuilt for some other
reason (e.g., cache reload). Also, it lets you easily see which
lists are being rebuilt in that case (as opposed to a situation where
all the lists fight over the global progress bar).
|
|
entry is destroyed.
This involves two things: referring to the entry indirectly via a slot,
so that progress calls after it's destroyed are ignored, and accepting
the entry as a reference, so that sigc::ref can be used to bind up weak
references when creating a callback to create the progress object.
|
|
set_progress() function in 2.16+.
|
|
object.
This paves the way for displaying the progress more flexibly.
|
|
|
|
I swear I made some of these changes in the last commit; I halfway
suspect that "hg record" glitched.
|
|
std::wstring.
This way we don't have to do several useless conversions to and from
UTF-8.
|
|
I feel like this makes things a little cluttered, but the short
descriptions are also very useful.
|
|
attached to a resolver tab.
This was causing icons and colors of items in a solution to fail to
update when the state of the items was changed (since the callbacks
that trigger this were connected to the global resolver, not the one
specific to the tab).
|
|
|
|
signals to its manager.
|
|
|
|
|
|
package. (Closes: #525898)
|
|
|
|
widget, rather than casting.
Hopefully this will reduce the temptation to do horribly wrong things
with memory. :-)
|
|
than casting the widget to a tab and then crashing.
|
|
This would have helped with tracking down two recent tab-close-related
bugs. In fact, I found the second one while instrumenting the code
with logging statements.
|
|
surprising situations where tabs with active method calls get closed.
This will avoid situations like the one we had with the resolver tab,
hopefully.
|
|
selection list and work on that.
This fixes the warnings that were getting spewed about how we were
modifying the list during iteration.
|
|
|
|
|
|
Before, the resolver would be automatically started sometimes even
though the user hadn't pushed the "find next solution" button. This
is now fixed: solutions won't be computed until the user is ready for
them (that's important so that their rejections/approvals are
reflected in any solution that gets generated).
|
|
|
|
boldface.
|
|
|
|
for action preference icons.
|
|
markup is used more consistent.
I'm not sure the current markup is what we want, but at least it's the
same everywhere and there are now fewer places to change if I decide to
rip it out.
|
