lyle / geekspeak.org Goto Github PK

"role" is an SQL-99 reserved keyword

Using lame you can add an image to the id3 info of an mp3...
Here is a sample:

lame --ti boxart.jpg --alt-preset cbr 64 -m m --resample 44.1 --lowpass 16.5 geekspeak\ 2016-05-18_mixdown.wav

Kinda a larger task. - maybe break this down to a few issues.

in the channel level description there are issues - use code for channel level itunes:summary for description as well

On the episode pages I added links to a few apps, these links are target=__blank - this causes them to not open their apps on mobile.
The ITunes one works on this page https://www.imafraidthat.com/new-blog/2018/10/31/episode14kids

this system puts html in description and not in itunes:summary - npr is using description and putting it into itunes:summary without stripping html. This might be causing SEO problems with itunes.

Episode statuses should be an enumerated type within the database. This will prevent typos from creeping into the DB and allow for the UI to be populated with enum values from the DB rather than hard coded in the app layer. Also takes up less disk space and can be queried faster.

e.g.

CREATE TYPE gs_status AS ENUM ('new', 'waiting', 'pending', 'live');

ALTER TABLE episodes
ALTER COLUMN status TYPE gs_status;

Extra Bullets are of issue. And Episode Date should be closer to show rather then leading it.

<itunes:author> is read at the episode level - adding the participants to this will help in SEO on iTunes.
Source: https://theaudacitytopodcast.com/podcast-seo-for-itunes-google-play-music-and-more-apps-tap291/

Using this schema https://github.com/RichardLitt/standard-readme

Because NPR seems to love to use the date they grab the episode from geekspeak.org as the "release date" - the end user does not know when the episode was first aired.

Also, a default teaser image would improve the look of a live episode.

On Feb 20 I moved geekspeak.org over to lets encrypt, but I neglected to update the capistrono scripts.
I setup a cron job (from https://www.digitalocean.com/community/tutorials/how-to-secure-nginx-with-let-s-encrypt-on-ubuntu-16-04) to get a renew every Monday.
The paths for the certs in the '/etc/nginx/site-enabled/geekspeak.org.conf' are updated, but the cap settings are not.
I also grabbed a copy of '/etc/letsencrypt' for safe keeping

When you search for GeekSpeak we are a good hit.
When you search for "Geek Speak" we are not getting a hit.
We want both to work.

User log data should use PostgreSQL "inet" type within the database. This will take up less space on disk as well as allow queries to run faster and be converted from IPv4 to IPv6 easily. Also prevents invalid data from getting into the database.

e.g.

ALTER TABLE users
ALTER COLUMN current_sign_in_ip TYPE inet,
ALTER COLUMN last_sign_in_ip TYPE inet;

remimber should be remember on the contact page. As per

Hi Geekspeak hosts,

there is a small typo on your contact page - http://geekspeak.org/contact

Instead of

"Please email us at [email protected] - and remimber to have a good
subject line. Most of us read every email."

it probably should read

"Please email us at [email protected] - and remember to have a good
subject line. Most of us read every email."

Love your show and listen to it frequently via the podcast.

Klaus [redacted]
Düsseldorf, Germany
Member of WHYY, Philadelphia

Amazon Echo's AI has trouble with the name GeekSpeak and does not seem to find the show. However, Geek Speak is not unique in the podcast space. Our Current title "GeekSpeak | How Technology Affects Society and Culture from the Geek Perspective" is way to long and was created to improve search results.
It is proposed, therefore, to change the name of the show to something simple and unique.

"Geek Speak with Lyle Troxell" seems to be a unique name and will probably allow it to be found easily via Voice AI.

When you edit the abstract and then click save an error message pops up asking if you want to discard changes - clicking cancel and then re-attempting save then saves successfully.

Two event handles are being added?

https://validator.w3.org/feed/check.cgi?url=http%3A%2F%2Fgeekspeak.org%2Fepisodes%2Frss.xml
More info:
http://support.libsyn.com/faqs/changing-your-feed-url-in-itunes/

Episode statuses should be an enumerated type within the database. This will prevent typos from creeping into the DB and allow for the UI to be populated with enum values from the DB rather than hard coded in the app layer. Also takes up less disk space and can be queried faster.

e.g.

CREATE TYPE gs_role AS ENUM ('new', 'waiting', 'pending', 'live');

ALTER TABLE participants
ALTER COLUMN role TYPE gs_role;

So that people can keep up its our show notes content create a non-podcast RSS feed.

Pull request: #24

When (not if) Rails dies, what is the procedure for restart?

There are a few typos on the new support page: Ocian, strugling, comforatable, formorly...

NPR is appending the station name to the front of the title:

KUSP-FM: GeekSpeak, KUSP Podcast

So we need to pull the ending ", KUSP Podcast" so that it will end up being:

KUSP-FM: GeekSpeak

I added a cron job that give the size of the mp3 files so we can determine how many downloads have taken place:
Here is the crontab entry for the splunk user
59 0 * * * ls -l /mnt/volume-sfo2-01/geekspeak-shared/shows/audio/ | awk '{if(NR==1){print "show_file,show_size";}else{print $9 "," int($5/1024) ;}}' > /opt/splunk/etc/apps/search/lookups/shows.csv

As per feedvalidator of feed going to iTunes....
the image title should match the main title.

Filter out all bits that are older then a month.

Some podcast catchers will display the Description as HTML in the player, allowing people to open links and such. If we change the item/description field to be the Abstract + the GeekBit Titles and Links we will have a much more useful show notes feature.