Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
tag: v0.8.0
Commits on Jul 19, 2012
  1. Properly report addfeed errors.

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Change URLs in manpages

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
  3. Remove extraneous output.

    authored
    We don't need to print the OPML file being imported...
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  4. Stub bugged/empty feeds with no previous items.

    authored
    The feed objects still exist, so we need to be able to count on shelf
    contents existing.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  5. Add basic manpages for daemon and remote

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Jul 18, 2012
  1. Py3 no longer should need to encode before printing

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Jul 17, 2012
  1. Remove extraneous hook logging.

    authored
    This can actually cripple hook performance depending on the number of
    registered hooks and their string representations ... even though the
    log level is generally too low to actually *use* the string...
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on May 13, 2012
  1. Hardent client code for multiple connections.

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on May 11, 2012
  1. Add Reddit plugin User-Agent

    authored
    Improve reddit plugin to report a User-Agent other than python's,
    because otherwise it's pegged as a script and somehow deprioritized /
    additionally ratelimited. You get HTTP 429 errors in this case.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Logging tweak.

    authored
    Strip NULL from the protocol stuff so that the log files are better
    behaved (i.e. detected as text files by programs like grep).
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Mar 5, 2012
  1. Trim unused tag configuration

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Mar 1, 2012
  1. Fix exception formatting.

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Feb 8, 2012
  1. Fix feedparser invocation in remote

    authored
    In particular, the code to grab the name of a feed automatically.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Fix reddit plugin

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Feb 7, 2012
  1. Update feedparser invocation.

    authored
    Fixes actual fetching with Py3
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Remove ancient alpha info from README

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
  3. Update README

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
  4. Further py3 fixes

    authored
    - errnos moved to e.args[0] from e[0]
    - remove unused sortfunc kwarg
    - strip useless decoder function
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  5. Initial port to Python 3

    authored
    There may be some lingering bugs in exception handlers.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Feb 6, 2012
  1. Further workaround fix

    authored
    Have to waitpid for that pid or we'll race. This also allows us to
    eliminate the CHLD handler.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Cleanup/fix workaround.

    authored
    Previous workaround was code with 'r' mode still. Fixed. Also, clean it
    up a bit by performing the open and close in the fork'd process.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  3. Workaround 13947

    authored
    Ugh, so the workaround in place for databases not shrinking themselves
    (thanks to bsddb disappearing), was to use the reorganize() function
    sporadically.
    
    This function has a bug I submitted: 13947 where it leaves file
    descriptors open that cause subsequent open() calls to fail with -EAGAIN
    until the process file descriptors are closed on exit().
    
    So, to workaround the bug in the workaround, we now fork(),
    reorganize(), and exit() in a pointless little temporary process.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Jan 26, 2012
  1. Properly check tag configs

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Reddit plugin v2

    authored
    This version abides by the request / 2 second limit Reddit imposes on
    clients.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  3. Add missing traceback import

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
  4. Make failed arg transforms return False, not None

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Dec 20, 2011
  1. Warn on type change, instead of ignoring.

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Dec 14, 2011
  1. Support automatically fetching attributes on ITEMS

    authored
    The current ATTRIBUTES command requires the client to know the ID of the
    story it's requesting attributes for, as it should. However, this means
    that two round trips have to be made when the IDs are unknown (ITEMS,
    then ATTRIBUTES).
    
    By adding the AUTOATTR command, the client can basically specify the
    attributes it knows it's going to need and the daemon will behave as if
    it received an ATTRIBUTES command for all of the returned items
    immediately afterwards.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Move protocol to JSON

    authored
    To improve interoperability, all external presence of the daemon is now
    JSON based. The config was changed awhile back, now the protocol is JSON
    parse-able as well.
    
    The major difference here is that the story ID is now a string, which is
    in turn a JSON string dump of a dict including the ID and the URL.
    Before, it was a (URL, ID) tuple, but that's not hashable and not
    guaranteed to be easily parsable elsewhere.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Dec 2, 2011
  1. Reorganize database with read-only permissions

    authored
    I'm not 100% sure how a reorg only requires read permissions. I'm clear
    on how the database only needs to be read, but then it's overwritten. I
    assume that's a separate permissions check.
    
    I do know one thing: if this is done as a writer, for some reason
    subsequent opens to the file return EAGAIN that will never resolve. The
    simplest explanation for that error is that the file descriptor is still
    open r/w (gdbm doesn't allow multiple writers for obvious reasons), but
    since the reorganization code immediately calls close() I'm not sure how
    that's possible.
    
    I'm also not convinced that this hasn't broken the reorganization code,
    but it doesn't except and I'll accept a slowly growing database over a
    crashing daemon any day.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Aug 31, 2011
  1. Fix c-r bad reference

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Aug 30, 2011
  1. Add --eval to c-r config to set values of types other than string.

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
  2. Proper solution for minor memory fix.

    authored
    Don't clear it, just make it fall out of scope right.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Aug 29, 2011
  1. Add ability for escplit to returned escaped terms.

    authored
    Signed-off-by: Jack Miller <jack@codezen.org>
Commits on Aug 26, 2011
  1. Ditch the hanging reference to old items.

    authored
    Decent memory improvement.
    
    Signed-off-by: Jack Miller <jack@codezen.org>
Something went wrong with that request. Please try again.