Tags

v0.5.2

Bugfix release 0.5.2.

Fixes are broken migrations (SQLite changed its behavior). This may have
left databases broken, but considering I'm probably the only user,
didn't implement automatically fixing them.

Also fixed a constraint SQLite couldn't actually enforce.

v0.5.1

Important bugfix release; non-parallel processing was broken.

Podist 0.5 broke non-parallel processing pretty badly. This release
fixes it.

Any playlists processed (non-parallel) with Podist 0.5 need to be
re-processed. Unfortunately, that feature isn't done yet (but is being
worked on).

v0.5

New feature release, Podist 0.5.

Major new features:

 1. Can now use librubberband to speed up (or slow down, if you really
    want) podcasts.

 2. Podcast processing (rubberband, volume leveling) can now be run in
    parallel. The default is not to, but testing here shows its
    benefitial up to around hyperthreads-1.

 3. Feature to drop podcasts. This will don't-use all remaining
    unplaylisted episodes and of course stop fetching new ones.
    Previously you had to do this directly in the database.

Minor new features:

  - Podist list -f now has a --long option to display a lot more data
    (requires very wide terminal).
  - "Podist editfeed" to edit existing feeds
  - "Podist editrandom" to edit random intermissions (e.g., change
    weight)
  - Can process podcasts and speech to MP3 CBR not just VBR. Apparently
    some players can't handle VBR.
  - More checks in Podist fsck.
  - Loudness normalization is now more accurate in make-random (really a
    bugfix...); was always this accurate in Podist proper.
  - make-random takes loudness target override on command line. Also has
    an option to re-process everything.
  - Will warn you if a random item was skipped for some weird reason
    (like not readable).

Removed things:

  - There used to be an undocumented and unlikely to be working
    "all_playlists" command. It is no more. (Long ago, it would generate
    as many playlists as possible then stop. Back when I was using this
    with a Neuros.)

Changes:

  - Default voice is now kal (rab, old default, required non-DFSG
    software).
  - Previously didn't have a <processing> section in the default config;
    now do.
  - Default target loudness is now -16 LUFS (was -23). A silly default,
    but, well, it's what everyone else has gone with, and now matches
    the volume of e.g., navigation on the phone. You can of course set
    whatever level you want in your config; this is merely the default.
  - Override default OpenSSL cipher list. Debian prohibited SHA-1, which
    broke some podcast downloads for me. Security isn't critical here,
    SHA-1 is fine (at least for now).

In addition, there have been a lot of bugfixes since 0.4.1, driven
largely by many new tests.

  - Creation of first playlist failed. Oops!
  - Don't call tput if not on a terminal (make-random).
  - fsck used to not be able to find some do-not-use files.
  - Work around perl bug 131192 (now at
    https://github.com/perl/perl5/issues/15959 apparently).

v0.4.1

Fix DB creation & upgrade; automated testing.

**Important bugfixes:** Creating a new database didn't work in v0.4.
Neither did upgrading from dbversion 1 databases (2–7 were OK, though).
(Note that dbversions don't match Podist version numbers; v0.4 uses
dbversion 8, for example)

**New features:** Now displays the feed number when adding a new feed,
so you can immediately pass that as "catch" (etc.)'s -f argument.

**Internals:** Not have GitLab CI running and have a few more tests;
that's how the above bugs were found, for example.

v0.4

v0.3

With the full implementation of limits on the number of items to fetch,
calling this Podist 0.3.