The more we try and patch up the fetcher code, the more things break. The
code blocks in question are practically unreadable and are full of corner
cases where fetching could fail. In summary the issues noticed included:
a) Always fetching strange broken urls from the premirror for "noclone"
git repositories
b) Not creating or rewriting .md5 stamp files inconsistently
c) Always fetching git source mirror tarballs from the premirror even
if they already exist but the checkout directory does now
d) Passing "None" values to os.access() and os.path.extsts() checks under
certain circumstances
e) Not using fetched git mirror tarballs if the preexist and always
try and fetch them.
This patch rewrites the sections of code in question to be simpler and
more readable, fixing the above problems and most likely other odd
corner cases.
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
We already export http,ftp,https proxy environment variables. Some environments,
GNOME for instance, place the socks proxy in ALL_PROXY and all_proxy. Export it
as well.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
We catch any exception raised by the fetchers go() method and attempt to work
around it by trying any (post) mirrors which are configured. However, should
the mirrors fail the exception is lost and the fetch is assumed to have
completed successfully.
Instead, save the exception and if the local file does not exist after trying
the mirrors re-raise the exception.
Fixes [BUGID #362]
Signed-off-by: Joshua Lock <josh@linux.intel.com>
With "no_proxy" exported to bitbake, both internal and externel file mirror can
be used. "https_proxy" enable fetching "https://" file through proxy.
Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Add a new method, try_premirror, to the Fetch object which checks to see
whether the file needs to be fetched from a premirror.
Override this in the Git fetcher to only require a pre-mirror fetch when the
clone directory does not exist.
Fixes [BUGID 290]
Signed-off-by: Joshua Lock <josh@linux.intel.com>
When setting localpath in the SRC_URI parameters the basename should be set
explicitly as the localpath() method is not called.
Fixes [BUGID #81]
Signed-off-by: Joshua Lock <josh@linux.intel.com>
If we fetch a tarball from a mirror it's entirely possible that the
mirror will have been from before the required tag/branch/etc was included
in the repository. To that end use forcefetch() as a way of testing whether
the repository is up to date and if not fetch updates.
Signed-off-by: Joshua Lock <josh@linux.intel.com>
When pulling from a premirror we would prefer a local tarball even when the
caller had specified the forcefetch parameter.
Add an extra parameter 'force' to try_mirrors, defaulting to False. If set
the mirrors will be tested even if the file exists locally.
Signed-off-by: Joshua Lock <josh@linux.intel.com>
* Fixes proxy support to honor standard proxy environment variables.
* Quote environment variables
(Bitbake rev: f84f382f340d6db15b9e5afb8c7c93969249a958)
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
Store database connection to persistent database in fetcher.
(Bitbake rev: 8a6876752b90efd81d92f0947bfc9527d8260969)
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
types.IntType -> int
types.StringType -> basestring
...
Also moves our ImmutableTypes tuple into our own namespace.
(Bitbake rev: 83674a3a5564ecb1f9d2c9b2d5b1eeb3c31272ab)
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
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.
(Bitbake rev: 3a11c2807972bbbddffde2fa67fc380d159da467)
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
Without this patch decoding a url of this kind file://dir/filename gives
path=/filename host=dir.
With the patch it decodes as path=/dir/filename host=""
Probably nobody stumbled on this issue yet because nobody used
file:// urls with directory names in the path.
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
When not expanding PREMIRRORS, the functions fails/does not work correctly
when PREMIRRORS is not a plain string (e.g. contains ${...} or a ${@...}
statements).
(Bitbake rev: d612d22b073f68b8cf1bb7344e0487820040d80d)
Signed-off-by: Enrico Scholz <enrico.scholz@informatik.tu-chemnitz.de>
Signed-off-by: Chris Larson <clarson@kergoth.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
dict objects provide an __iter__ method for the iteration which gives you the
keys, so calling keys directly is unnecessary, and isn't really a best
practice. The only time you really need to call the keys is if there's a
danger of the dict changing out from underneith you, either due to external
forces or due to modification of the iterable in the loop. Iterations over
os.environ are apparently subject to such changes, so they must continue to
use keys().
As an aside, also switches a couple spots to using sorted() rather than
creating a temporary list with keys() and sorting that.
(Bitbake rev: 5b6ccb16c6e71e23dac6920cd2df994d67c2587b)
Signed-off-by: Chris Larson <clarson@mvista.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
(Bitbake rev: f68406e864c9837feb56cbec993b620481445cc2)
Signed-off-by: Tom Rini <tom_rini@mentor.com>
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>
checkstatus() is used to ensure we can fetch a copy of each file, so it makes
sense to also test PREMIRRORS and MIRRORS in the method.
This patch adds calls to try_mirrors() to the Fetch.checkstatus() method and
changes the try_mirrors() method to take a check argument, which is False by
default. When check is True try_mirrors() will call a fetchers checkstatus()
with the replaced uri.
Signed-off-by: Joshua Lock <josh@linux.intel.com>
The logic in try_mirrors() was buggy such that only the first entry in the
mirrors list was tried.
This patch fixes this. Thanks to Richard for the review.
Signed-off-by: Joshua Lock <josh@linux.intel.com>
When we are using PREMIRRORS it's possible a mirror in the local namespace
(some filesystem path, i.e. an NFS share) provides read-only files.
This is a perfectly valid scenario so this patch fixes bitbake so that for
such a scenario locapath is set to the files path rather than some child
of DL_DIR.
Signed-off-by: Joshua Lock <josh@linux.intel.com>
Modify the try_mirrors() function to return the localpath of the fetched file
and update the data dictionary to reflect this.
Secondly the metadata files, lock and md5, should always be stored relative to
the ${DL_DIR} as it is possible that the localpath is a read-only directory,
for example in the scenario where there is a read-only file:// mirror.
Signed-off-by: Joshua Lock <josh@linux.intel.com>
This patch serves two purposes. Firstly it unifies the concept of mirrors into
PREMIRRORS and MIRRORS. PREMIRRORS are tried before the SRC_URI defined in the
recipe whereas MIRRORS are tried only if that fails.
The tarball stash was conceptually inline with a PREMIRROR only with special
handling within the wget fetcher and therefore only worked with certain
fetch types.
Secondly the patch removes the need for individual fetch implementations to
worry about mirror handling.
With this patch, the base fetch implementation will first try to use a
PREMIRROR to fetch the desired object, if this fails the native fetch method
for the object will be tried and if this fails will try to fetch a copy from
one of the MIRRORS.
Signed-off-by: Joshua Lock <josh@linux.intel.com>
