anaconda-patches March 2013

anaconda-patches@lists.fedorahosted.org

13 participants
90 discussions

[master][PATCH] Check _current_action for not being None before using it
by Vratislav Podzimek 11 Mar '13

11 Mar '13

It may happen that _current_action is None even when self._actions is not None and not empty. --- pyanaconda/ui/gui/__init__.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/pyanaconda/ui/gui/__init__.py b/pyanaconda/ui/gui/__init__.py index 1e432d4..7ace103 100644 --- a/pyanaconda/ui/gui/__init__.py +++ b/pyanaconda/ui/gui/__init__.py @@ -309,7 +309,7 @@ class GraphicalUserInterface(UserInterface): from gi.repository import AnacondaWidgets # if there are no actions (not populated yet), we can do nothing - if len(self._actions) > 0: + if len(self._actions) > 0 and self._currentAction: lightbox = AnacondaWidgets.lb_show_over(self._currentAction.window) window.main_window.set_transient_for(lightbox) -- 1.7.11.7

2 1

[PATCH] Bonding support - GUI
by Radek Vykydal 11 Mar '13

11 Mar '13

TBD: - sync ifcfgs for initscripts and NM (https://bugzilla.redhat.com/show_bug.cgi?id=919001) - (?) removing bond device with [-] button

3 14

[master] Patch for the new libxklavier
by Vratislav Podzimek 08 Mar '13

08 Mar '13

This adapts our XklWrapper to the new behaviour of libxklavier. Though it seems to work I'm not sure it will work correctly with some non-ascii strings. Maybe they are gone now. I'll ask the libxklavier maintainer what's going on, but let's start with this patch making rawhide Anaconda work again. Vratislav Podzimek (1): Adapt to the new libxklavier's behaviour pyanaconda/keyboard.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) -- 1.7.11.7

2 4

[PATCH] Fix two places where we are locking up the main thread (#886680).
by Chris Lumens 07 Mar '13

07 Mar '13

(1) Letting _first_refresh be the first time we reference payload.groups and payload.environments means we are potentially downloading lots of big data in the main thread, which means the UI cannot update. (2) Referencing anything in software.py's _initialize grabs the yum lock, possibly for a long time. When we later check the source spoke's status, we reference ready which checks baseRepo, which also attempts to grab the same lock. This isn't a deadlock because it will eventually get released, but it does make the UI unresponsive for a while. --- pyanaconda/ui/gui/spokes/software.py | 24 ++++++++++++++++++------ pyanaconda/ui/gui/spokes/source.py | 7 +++---- 2 files changed, 21 insertions(+), 10 deletions(-) diff --git a/pyanaconda/ui/gui/spokes/software.py b/pyanaconda/ui/gui/spokes/software.py index bc21af9..14b15c6 100644 --- a/pyanaconda/ui/gui/spokes/software.py +++ b/pyanaconda/ui/gui/spokes/software.py @@ -24,6 +24,7 @@ _ = lambda x: gettext.ldgettext("anaconda", x) N_ = lambda x: x from pyanaconda.flags import flags +from pyanaconda.packaging import MetadataError from pyanaconda.threads import threadMgr, AnacondaThread from pyanaconda.ui import communication @@ -184,8 +185,25 @@ class SoftwareSelectionSpoke(NormalSpoke): # We don't want to do a full refresh, just join the metadata thread threadMgr.wait("AnaPayloadMDThread") else: + # Grabbing the list of groups could potentially take a long time + # at first (yum does a lot of magic property stuff, some of which + # involves side effects like network access. We need to reference + # them here, outside of the main thread, to not block the UI. + try: + self.payload.environments + self.payload.groups + except MetadataError: + communication.send_message(self.__class__.__name__, + _("No installation source available")) + return + + # And then having done all that slow downloading, we need to do the first + # refresh of the UI here so there's an environment selected by default. + # This happens inside the main thread by necessity. We can't do anything + # that takes any real amount of time, or it'll block the UI from updating. if not self._first_refresh(): return + self.payload.release() communication.send_ready(self.__class__.__name__) @@ -196,12 +214,6 @@ class SoftwareSelectionSpoke(NormalSpoke): @gtk_thread_wait def _first_refresh(self): - # Grabbing the list of groups could potentially take a long time the - # first time (yum does a lot of magic property stuff, some of which - # involves side effects like network access) so go ahead and grab - # them once now. - from pyanaconda.packaging import MetadataError - try: self.refresh() return True diff --git a/pyanaconda/ui/gui/spokes/source.py b/pyanaconda/ui/gui/spokes/source.py index 4b3d90d..6381130 100644 --- a/pyanaconda/ui/gui/spokes/source.py +++ b/pyanaconda/ui/gui/spokes/source.py @@ -614,10 +614,9 @@ class SourceSpoke(NormalSpoke): @property def ready(self): from pyanaconda.threads import threadMgr - # By default, the source spoke is not ready. We have to wait until - # storageInitialize is done to know whether or not there's local - # devices potentially holding install media. - return (self._ready and not threadMgr.get("AnaPayloadMDThread") and + return (self._ready and + not threadMgr.get("AnaPayloadMDThread") and + not threadMgr.get("AnaSoftwareWatcher") and not threadMgr.get("AnaCheckSoftwareThread")) @property -- 1.8.1.2

2 1

[PATCH] Log when we acquire and release the _yum_lock (dlehman, clumens).
by Chris Lumens 07 Mar '13

07 Mar '13

We've got a temporary locking issue somewhere that causes the UI to become unresponsive for a bit (see #886680, perhaps other bugs). Examining the code indicates this isn't necessarily due to running things in the main loop that shouldn't be. There's something waiting for the _yum_lock. This patch adds logging to help debug. --- pyanaconda/packaging/yumpayload.py | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) diff --git a/pyanaconda/packaging/yumpayload.py b/pyanaconda/packaging/yumpayload.py index 9549420..fe132f7 100644 --- a/pyanaconda/packaging/yumpayload.py +++ b/pyanaconda/packaging/yumpayload.py @@ -94,9 +94,25 @@ from pykickstart.constants import KS_MISSING_IGNORE default_repos = [productName.lower(), "rawhide"] -from threading import RLock -_yum_lock = RLock() +import inspect +import threading +_private_yum_lock = threading.RLock() +class YumLock(object): + def __enter__(self): + frame = inspect.stack()[2] + threadName = threading.currentThread().name + + log.info("about to acquire _yum_lock for %s at %s:%s (%s)" % (threadName, frame[1], frame[2], frame[3])) + _private_yum_lock.acquire() + log.info("have _yum_lock for %s" % threadName) + return _private_yum_lock + + def __exit__(self, exc_type, exc_val, exc_tb): + _private_yum_lock.release() + log.info("gave up _yum_lock for %s" % threading.currentThread().name) + +_yum_lock = YumLock() _yum_cache_dir = "/tmp/yum.cache" class YumPayload(PackagePayload): -- 1.8.1.2

2 2

[master] [PATCH] Redraw screen in case of valid input and nothing new scheduled
by Vratislav Podzimek 06 Mar '13

06 Mar '13

If the screen successfully processes the input and doesn't schedule another screen, it should be redrawn, so that the new state is shown. Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com> --- pyanaconda/ui/tui/simpleline/base.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/pyanaconda/ui/tui/simpleline/base.py b/pyanaconda/ui/tui/simpleline/base.py index 5e2e8be..131a54a 100644 --- a/pyanaconda/ui/tui/simpleline/base.py +++ b/pyanaconda/ui/tui/simpleline/base.py @@ -269,7 +269,7 @@ class App(object): except Exception as ex: send_exception(self.queue, ex) return False - + else: # this can happen only in case there was invalid input and prompt # should be shown again @@ -341,6 +341,10 @@ class App(object): # increment the error counter if not self.input(self._screens[-1][1], c): error_counter += 1 + else: + # input was successfully processed, but no other screen was + # scheduled, just redraw the screen to display current state + self.redraw() # redraw the screen after 5 bad inputs if error_counter >= 5: -- 1.7.11.7

2 2

[master] Two exception handling related patches
by Vratislav Podzimek 06 Mar '13

06 Mar '13

PATCH 1/2 uses the proposed patch for python-meh to replace magic tuples with much clearer named tuples and may be squashed with the patch ensuring exception handling in text mode runs in the main thread. PATCH 2/2 fixes bug nobody has probably hit yet. Vratislav Podzimek (2): Use namedtuple instead of our magic tuples We use python-meh's interfaces instead of Anaconda's pyanaconda/exception.py | 35 +++++++++++++++++------------------ 1 file changed, 17 insertions(+), 18 deletions(-) -- 1.7.11.7

2 3

[python-meh][PATCH] Use named tuples instead of our magic tuples
by Vratislav Podzimek 06 Mar '13

06 Mar '13

Using ExceptionInfo and DumpInfo named tuples makes code clearer and simplifies documentation. Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com> --- meh/__init__.py | 8 ++++++++ meh/dump.py | 24 ++++++++++++------------ meh/handler.py | 54 +++++++++++++++++++++++++++--------------------------- po/python-meh.pot | 2 +- tests/baseclass.py | 3 ++- 5 files changed, 50 insertions(+), 41 deletions(-) diff --git a/meh/__init__.py b/meh/__init__.py index 329969f..d09a131 100644 --- a/meh/__init__.py +++ b/meh/__init__.py @@ -20,6 +20,8 @@ # Author(s): Chris Lumens <clumens(a)redhat.com> # +from collections import namedtuple + # These constants represent the return values of buttons on the initial # exception handling dialog - the dialog that first pops up when an # exception is hit. @@ -127,3 +129,9 @@ class Config(object): else: msg = "Callback with name '%s' already registered" % item_name raise ConfigError(msg) + +# type, value and stack are items provided by Python for the exception handler +ExceptionInfo = namedtuple("ExceptionInfo", ["type", "value", "stack"]) + +# ExceptionInfo instance plus the object that should be dumped +DumpInfo = namedtuple("DumpInfo", ["exc_info", "object"]) diff --git a/meh/dump.py b/meh/dump.py index c868aa4..4639327 100644 --- a/meh/dump.py +++ b/meh/dump.py @@ -35,23 +35,23 @@ class ExceptionDump(object): for manipulating a traceback. In general, clients should not have to use this class. It is mainly for internal use. """ - def __init__(self, (ty, value, stack), configObj): + def __init__(self, exc_info, config_obj): """Create a new ExceptionDump instance. Instance attributes: - (type, value, stack) -- These are the values provided by python - when an exception is generated, and are - what ExceptionDump wraps to make more - useful. - conf -- An instance of the Config object. + @param exc_info: info about the exception provided by Python + @type exc_info: an instance of the meh.ExceptionInfo class + @param config_obj: configuration for python-meh + @type config_obj: an instance of the meh.Config class """ - if inspect.istraceback(stack): - stack = inspect.getinnerframes(stack) + if inspect.istraceback(exc_info.stack): + self.stack = inspect.getinnerframes(exc_info.stack) + else: + self.stack = exc_info.stack - self.conf = configObj - self.stack = stack - self.type = ty - self.value = value + self.conf = config_obj + self.type = exc_info.type + self.value = exc_info.value self._dumpHash = {} diff --git a/meh/handler.py b/meh/handler.py index 6338cf2..6427656 100644 --- a/meh/handler.py +++ b/meh/handler.py @@ -68,7 +68,7 @@ class ExceptionHandler(object): exn = property(lambda s: s._exn, lambda s, v: s._setExn(v)) - def handleException(self, (ty, value, tb), obj): + def handleException(self, dump_info): """This is the main exception handling entry point. When Python gets an exception it doesn't know how to handle, this method will be called. It then saves the traceback and displays the main @@ -79,6 +79,9 @@ class ExceptionHandler(object): All arguments are passed in from the handler created by calling self.install(). This method should not usually be overridden by a subclass. + + @type dump_info: an instance of the meh.DumpInfo class + """ responseHash = {MAIN_RESPONSE_QUIT: self.runQuit, @@ -86,34 +89,34 @@ class ExceptionHandler(object): MAIN_RESPONSE_SAVE: self.runSave} # Quit if we got an exception when running pdb. - if isinstance(value, bdb.BdbQuit): + if isinstance(dump_info.exc_info.value, bdb.BdbQuit): sys.exit(self.exitcode) # Restore original exception handler. sys.excepthook = sys.__excepthook__ - self.preWriteHook((ty, value, tb), obj) + self.preWriteHook(dump_info) # Save the exception to the filesystem first. - self.exn = self.exnClass((ty, value, tb), self.conf) + self.exn = self.exnClass(dump_info.exc_info, self.conf) (fobj, self.exnFile) = self.openFile() - self.exnText = self.exn.traceback_and_object_dump(obj) + self.exnText = self.exn.traceback_and_object_dump(dump_info.object) fobj.write(self.exnText) fobj.close() - self.postWriteHook((ty, value, tb), obj) + self.postWriteHook(dump_info) # Run initial UI screen, asking whether to save/debug/quit. while True: win = self.intf.mainExceptionWindow(str(self.exn), self.exnText) if not win: - self.runQuit((ty, value, tb)) + self.runQuit(dump_info.exc_info) rc = win.run() win.destroy() try: - responseHash[rc]((ty, value, tb)) + responseHash[rc](dump_info.exc_info) except KeyError: # This happens if (for example) the user hits Escape instead # of pressing a button. @@ -121,33 +124,23 @@ class ExceptionHandler(object): win.destroy() - def preWriteHook(self, (ty, value, tb), obj): + def preWriteHook(self, dump_info): """Subclasses may supply a function with this name that will be called immediately before the traceback is written to disk in order to have any sort of special pre-write processing that needs to be done. - (ty, value, tb) -- The Python objects created when a traceback - occurs. These are passed in directly from - handleException. - obj -- A Python object that may be dumped to a file when the - exception is saved. This should be something like the top - level object in a program. + @type dump_info: an instance of the meh.DumpInfo class """ pass - def postWriteHook(self, (ty, value, tb), obj): + def postWriteHook(self, dump_info): """Subclasses may supply a function with this name that will be called immediately after the traceback is written to disk, but immediately before the UI is run. This is to provide a place for any special handling to happen once there is a file on disk. - (ty, value, tb) -- The Python objects created when a traceback - occurs. These are passed in directly from - handleException. - obj -- A Python object that may be dumped to a file when the - exception is saved. This should be something like the top - level object in a program. + @type dump_info: an instance of the meh.DumpInfo class """ pass @@ -160,7 +153,8 @@ class ExceptionHandler(object): exception is saved. This should be something like the top level object in a program. """ - sys.excepthook = lambda ty, value, tb: self.handleException((ty, value, tb), obj) + sys.excepthook = lambda ty, value, tb: \ + self.handleException(DumpInfo(ExceptionInfo(ty, value, tb), obj)) def openFile(self): """Create a randomly named output file to write the exception dump to. @@ -176,29 +170,35 @@ class ExceptionHandler(object): ### Methods called by the primary exception handling dialog. - def runQuit(self, (ty, value, tb)): + def runQuit(self, exc_info): """This method is called when the "Exit" button is clicked. It may be overridden by a subclass, but the basic behavior of eventually quitting should be preserved. + + @type exc_info: an instance of the meh.ExceptionInfo class """ sys.exit(self.exitcode) - def runDebug(self, (ty, value, tb)): + def runDebug(self, exc_info): """This method is called when the "Debug" button is clicked. It may be overridden by a subclass if specialized behavior is required to enter debug mode. + + @type exc_info: an instance of the meh.ExceptionInfo class """ print print "Use 'continue' command to quit the debugger and get back to "\ "the main menu" import pdb - pdb.post_mortem(tb) + pdb.post_mortem(exc_info.stack) #no need to quit here, let's just get back to the main dialog - def runSave(self, (ty, value, tb)): + def runSave(self, exc_info): """This method is called when the "Save" button is clicked. It may be overridden by a subclass, but that's going to be a lot of work. + + @type exc_info: an instance of the meh.ExceptionInfo class """ params = dict() diff --git a/po/python-meh.pot b/po/python-meh.pot index 44395da..5d4f33f 100644 --- a/po/python-meh.pot +++ b/po/python-meh.pot @@ -8,7 +8,7 @@ msgid "" msgstr "" "Project-Id-Version: PACKAGE VERSION\n" "Report-Msgid-Bugs-To: \n" -"POT-Creation-Date: 2013-01-21 12:43+0100\n" +"POT-Creation-Date: 2013-03-05 12:14+0100\n" "PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n" "Last-Translator: FULL NAME <EMAIL@ADDRESS>\n" "Language-Team: LANGUAGE <LL(a)li.org>\n" diff --git a/tests/baseclass.py b/tests/baseclass.py index a0173b7..befa9b4 100644 --- a/tests/baseclass.py +++ b/tests/baseclass.py @@ -5,6 +5,7 @@ import sys import tempfile import unittest +from meh import ExceptionInfo from meh.handler import * from meh.dump import * @@ -13,7 +14,7 @@ class BaseTestCase(unittest.TestCase): from inspect import stack as _stack stack = _stack()[1:] - dump = ExceptionDump((None, None, stack), conf) + dump = ExceptionDump(ExceptionInfo(None, None, stack), conf) return dump.dump(obj) def openFile(self): -- 1.7.11.7

2 2

[PATCH] Could not load UI file advanced_user.glade
by Mark Hamzy 06 Mar '13

06 Mar '13

Fixes IOError: Could not load UI file 'advanced_user.glade' for object --- pyanaconda/ui/gui/spokes/user.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/pyanaconda/ui/gui/spokes/user.py b/pyanaconda/ui/gui/spokes/user.py index 93b4d08..af39856 100644 --- a/pyanaconda/ui/gui/spokes/user.py +++ b/pyanaconda/ui/gui/spokes/user.py @@ -56,7 +56,7 @@ def strip_accents(s): class AdvancedUserDialog(GUIObject): builderObjects = ["advancedUserDialog", "uid", "gid"] mainWidgetName = "advancedUserDialog" - uiFile = "advanced_user.glade" + uiFile = "spokes/advanced_user.glade" def __init__(self, user, groupDict, data): GUIObject.__init__(self, data) -- 1.8.1.2

2 1

[PATCH] Ignore MTDs, as we do not have the tools to write to them (#916771).
by Chris Lumens 04 Mar '13

04 Mar '13

--- blivet/devicetree.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/blivet/devicetree.py b/blivet/devicetree.py index 88ba130..c154343 100644 --- a/blivet/devicetree.py +++ b/blivet/devicetree.py @@ -501,6 +501,10 @@ class DeviceTree(object): if name.startswith("ram"): return True + # Memory Technology Devices require special tools to manipulate. + if name.startswith("mtd"): + return True + if name.startswith("loop"): # ignore loop devices unless they're backed by a file return (not devicelibs.loop.get_backing_file(name)) -- 1.8.1.2

2 1

← Newer
1
2
3
4
5
6
7
8
9
Older →

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

anaconda-patches March 2013