Fix syntax error in shell.py

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add missing 'return False' to the SystemExit handler in runAsyncCommand

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Move serverConnection.terminate() back into the finally

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Limit the traceback length in the default exception handler

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Switch from our own 'dummywrite' class to StringIO

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Switch a fatal usage to SystemExit ("T" not set)

This failure is one the *user* needs to see, as it may have to do with their
setup, but where they don't need to see a traceback and all, since they're not
developers, so I think SystemExit is appropriate here.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Resurrect the excepthook, this time just sending the info with bb.fatal

Signed-off-by: Chris Larson <chris_larson@mentor.com>

remove return in finally statement

causes exceptions to not be handled after the finally statement

Signed-off-by: Cliff Brake <cbrake@bec-systems.com>
Signed-off-by: Chris Larson <clarson@kergoth.com>

cooker: Kill __getitem__ usage on exception objects (python 3 prep)

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Handle SystemExit and KeyboardInterrupt sanely when executing a command

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop unnecessarly duplicated messages wrt FuncFailed

Signed-off-by: Chris Larson <chris_larson@mentor.com>

knotty: use the future division to prep for 3.x

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Kill the unnecessary setVarFlag debugging message

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add exception bits to TODO

Signed-off-by: Chris Larson <chris_larson@mentor.com>

knotty: stop freaking out and dying when bb.build has removed the empty logfile

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Use bb.utils.mkdirhier in bb.cache to shut up the pending deprecation warning

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Use PendingDeprecationWarning for functions moved from bb to bb.utils

This is necessary, as there has not yet been a release with the bb.utils
versions in place.  We can't show them a deprecation warning when they can't
safely change it yet.  Show a PendingDeprecationWarning until 1.10 releases
and OE requires it, then we can move to DeprecationWarning.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add fatal bits and LAYERDIR bits to TODO

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Start ditching fatal() calls in favor of raising appropriate exceptions

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop the non-debug traceback suppressor, at this point I think it does more harm than good

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Change wget fetcher to use the runfetchcmd

* Fixes proxy support to honor standard proxy environment variables.
* Quote environment variables

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Show a traceback when a task fails due to an exception

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Update the TODO

Signed-off-by: Chris Larson <chris_larson@mentor.com>

parser: fix harmless typo in variable name

s/absolsute/absolute/

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

Fix the silent exit when the specified task doesn't exist

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Move the output of no provider and multiple provider messages into the UI

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Reorganize the main exception handling so an attribute exception raised in the ui isn't misinterpreted

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Make knotty ignore the NoProvider event

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Don't tell the user we're sending SIGINT to the remaining 0 tasks

Signed-off-by: Chris Larson <chris_larson@mentor.com>

cooker: use time.strftime() instead of spawing sh with date

time.strftime() defaults to using time.localtime(), use that instead
of spawning "sh -c 'date +<FMT>'"

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

Adjust the exception handling based on Cliff Brake's patch

Rather than printing the exception ourselves, might as well re-raise and let
the default exception handler do it for us.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Fix a couple typos in the CVS Fetcher section.

"DL_DIRS" -> "DL_DIR"
Remove superfluous trailing ">" after CVSDIR.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

Fix a couple typos in setup.py.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

cache: do not chdir unnecessarily

previously we called chdir() twice for every target, this patch reduces
the amount of chdir() calls via openembedded master from some 16000 to
4.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

persist_data: cache connection and use cursor

Store database connection to persistent database in fetcher.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

cache: use a set() for __depends

to make updating depends easier/more intuitive/eventually faster

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

cache: use max() to determine newest_mtime

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

cache: use os.path ops instead of stat

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

Make the file not found error actually useable.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

fetch: commentary typo fix in fetcher_init()

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

cache: fix typo in error message of cacheValidUpdate()

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

fetch: documentation typo fix

s/initilize/initialize/

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

fetch: Check for existence of tar file along with md5 signature

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

Revert "In expand, drop the unnecessary second regular expression match"

This reverts commit 05462fa7908fc22988b3dc9d376798d0a46ccb5a.

Don't emit vars starting with __, as those are internal

Signed-off-by: Chris Larson <chris_larson@mentor.com>

In expand, drop the unnecessary second regular expression match

Signed-off-by: Imran Mehmood <imran_mehmood@mentor.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>

Simplify _findVar a bit

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Fix major bug that slipped in when moving update_data

The root cause is, I was testing the use of renameVar() from finalize, in
order to get flags copied over when an override is applied, but renameVar
removes the original, whereas the old code did not do so.  Going back to the
old method, will revisit the override/flags later on.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Move a var definition in cooker.parse_next

Signed-off-by: Chris Larson <clarson@kergoth.com>

Enhance the expansion exception message

Signed-off-by: Chris Larson <chris_larson@mentor.com>

For ${@}, run str() on the result

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Allow emission of non-string objects in emit_var, instead running str() on it first

Signed-off-by: Chris Larson <chris_larson@mentor.com>

emit_env: clean up, iterate once

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Fix the debug level check in print_exception

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Don't try to expand non-string values

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Move the python version check into the python package

This ensures the check is run for anyone using the package, including
setup.py, though it's sad that it's necessary, since all it really wants is
__version__.

Ideally, we'd avoid the version check entirely in favor of checking for
functionality, but that's rather difficult with language constructs like
context managers, so this'll do for now.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop an unnecessary second python version check and use the less ugly sys.version_info

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Bump version to 1.11.0

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Move the python version check before the module imports, in case the modules utilize the newer python features

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Search up the directory tree for bblayers.conf

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add a hack to avoid the req that all vars using LAYERDIR be immediately expanded

See the comment in the code for details.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Move the unset BBPATH sanity check after the bblayers processing

This way we can fully utilize bblayers, you can do everything in bblayers.conf
and avoid setting any environment variables at all.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Kill unnecessary usages of the types module

types.IntType -> int
types.StringType -> basestring
...

Also moves our ImmutableTypes tuple into our own namespace.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Apply some 2to3 transforms that don't cause issues in 2.6

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add __contains__ to COWDictMeta

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop unused bb.data.expandData function

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Ignore deprecation warnings for now

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Switch some references to moved functions

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add a missing import of print_function

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Apply the 2to3 print function transform

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Deprecate the usage of certain objects via certain modules

As an example, this displays a deprecation warning for the use of
"bb.encodeurl" when you should be using "bb.fetch.encodeurl".  It includes a
convenience function for this purpose.  It should be of use when moving
objects between modules permanently, changing the API the user sees.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Exception handling fixups

Add some missing docstrings

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop __all__ usage

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop some unused functions

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop some unused classes

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop some unused vars

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Import fixups

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Formatting cleanups

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Don't run metadata event handlers for bb.msg events

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Stop using functions via 'bb' that were moved from there to other modules

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add a function to get the debug level, without poking into debug_levels manually

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Fix invalid logging domain warning

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Bump minimum python version check to 2.6.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Make the default domain a proper domain, not a string

This lets you do -l Default on the commandline to bump the debug level of
messages with no domain, without bumping the level of everything else.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Implement bb.msg.domain as a named tuple, drop the Enum class

Also fixes some bb.msg references from within bb.msg.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Fix an error in the ncurses UI

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Display python warnings via bb.msg

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Immediately display messages if no UI handlers are installed yet

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Use __import__ for the ui rather than exec

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Fix the task event handling in the ncurses UI

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Ensure we always utilize the correct messaging domains

Signed-off-by: Chris Larson <chris_larson@mentor.com>

bb.event: NotHandled and Handled are on the way out

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Rename the finalise method for consistency

Apparently the finalise spelling is becoming less common in British English.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Move update_data into the DataSmart class as a finalize() method

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Initialize pkgs_to_build in the BBConfiguration constructor

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Drop doctest, as we aren't utilizing it

We should add back in some proper unit testing.  Doctest is rather limited,
and we haven't actually made use of it since the original implementation of
the datastore when the project started, as far as I'm aware.

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Add bin/bitbakec to .gitignore

Signed-off-by: Chris Larson <chris_larson@mentor.com>

Made '-b' work with BBCLASSEXTEND

When BBCLASSEXTEND is set, '-b' builds usually failed with messages
like

| ERROR: Parsing error data_fn virtual:native:<recipe>.bb and fn
<recipe>.bb don't match
|
|   File ".../bb/providers.py", line 47, in sortPriorities
|     priority = dataCache.bbfile_priority[f]
| KeyError: 'virtual:native:<recipe>.bb'

This patch fixes it and allows to specify the alternative class in a
way like

| ./bitbake -b virtual:native:<recipe>.bb

This patch was written to be so minimal as possible; variables should be
probably renamed to reflect their new meaning.

Signed-off-by: Enrico Scholz <enrico.scholz@informatik.tu-chemnitz.de>

Resurrect the old bb.vercmp as bb.utils.vercmp, and its deps

This is just for compatibility.  We may drop it in the future, or rewrite it,
as it's not particularly pythonic.

Signed-off-by: Chris Larson <chris_larson@mentor.com>