distcc 3.3.2

I’ve finally updated distcc to the latest release (f31+). Python 2 retirement forced me to finally work out how to adapt to the new compiler symlink model. It should work out of the box, but if you add compilers to a server, rerun update-distcc-symlinks.


Two numpys diverged in a wood. . .

python2-numpy has split off from numpy(python3-numpy) as of rawhide/f31.  Dependencies have remained the same, so unless your Python 2 package requires numpy, you should be fine. If it does, switch to python2-numpy. Or switch to python3. 🙂

This allows us to upgrade python3-numpy to 1.17.0, while python2-numpy stays at 1.16.4, since 1.17.x drops Python 2 support.

File bugs if something breaks.

Duplicity 0.8.01

Duplicity 0.8.01 is now in rawhide. The big change here is that it now uses Python 3. I’ve tested it in my own environment, both on it’s own and with deja-dup, and both work.

Please test and file bugs. I expect there will be more, but with Python 2 reaching EOL soon, it’s important to move everything we can to Python 3.



Our long multimedia nightmare is over

So, for quite some time, I maintained mediatomb. I didn’t use it, but I kept it working for those who did.

As time passed, it more or less died upstream, and was replaced by Gerbera. So, naturally, I packaged it, and its dependencies, got the reviews in, etc. I retired mediatomb, and off we went. As of Fedora 29, there is no mediatomb, only Gerbera.

One minor issue.

With the initial version, I shipped a config based on mediatomb’s config. This didn’t work, because Gerbera’s config format is completely different.

I fixed it. 🙂 If you like, test and give karma.

As a side note, I now use Gerbera. It’s pretty cool.

Fedora 29 Beta

It’s Beta time again!

As is my habit, I upgraded my laptop at Beta time. dnf system-upgrade didn’t work for me because of some dependency issues. In the process of working through a dnf upgrade, I discovered that it was due to some odd homegrown Python RPMs I’d made and forgotten about, and gource, which was still FBTBS. After working those out, it was uneventful.

That was last night. Now I’ve been using it for several hours.

It’s still uneventful.

Great job, Fedora folx, I think GA should be smooth. (knock on wood)

Joe macros FTW

I’ve been using Linux for more years than I care to admit, but it’s been long enough that my choice of text editor was informed by what was available for the first Linux distribution I tried that:

  1. Had a gentler learning curve than vi/vim
  2. Had a binary package that fit on a single floppy (so no emacs)

Thus began my multi-decade love affair with Joe.

Fast forward to the beginning of my time editing RPM spec files; first for myself, then for Fedora Extras, then for Fedora at large, EPEL, employers, etc. One of the most common errors I see are bogus dates in changelogs. That, and people who manage to misspell their own names and/or email addresses. I’ve done it. If you’ve made any reasonable number of spec edits, you’ve almost certainly done it. So we’re careful. We check the date, the day of the week, and do our best. We copy/paste/edit when we can.

I shudder to think of the time I’ve lost taking care to get my changelog entries right.

This morning, I finally worked out a time-saver, and I think some of you might find it useful.

vi, emacs, and presumably other editors, allow you to insert the output of a script into the current buffer at the location of the cursor. Joe will only append to the end of the file, which is silly, and if someone knows how to insert at the current location, please let me know.

But at any rate, the following bash script:

DATE=`date +%a\ %b\ %d\ %Y`
NAME=`git config --get user.name`
EMAIL=`git config --get user.email`
echo "* $DATE $NAME <$EMAIL>"

will output a usable string as follows:

* Wed Jul 25 2018 Your Name <youremail@example.com>

For joe, you need to pipe this to a file and then create a macro like:

Ctrl+K [ (choose a number
Ctrl+K ]

And then run it with:

Crtl+k <number>

And you’re set!