Nectarine

Found me a search bug:
https://scenestream.net/search/?q=Hyperlucky
which finds nothing, so I was fixing to upload it.

But of course it's right here:
https://scenestream.net/demovibes/song/45789/
which I found by searching for its companion piece, Plastic Dove.

I tried just "lucky" with the "songs only" tick box, and also got nothing. This is a recent 2019 song so I would guess it relates to something being out of date.
So does the search operate on a special "search cache" rather than the actual current list of songs?
But it's an extra bad bug because it still takes a fucking week and you think "yep it's definitely checking everything!" but the joke's on you! The song was there all along, and the search was actually away drinking a beer and playing Tetris.

Unfortunately, search indexing is broken (and has been for some time). But only on the production server for some reason. It works great in Dev.

Once a long time ago I tried to to fix it, but the root cause was a circular dependency in a couple of the Python modules. And when I fixed that, the site started throwing 500 errors all the time instead. So I put things back as they were.

Notifications seem to not disappear until I press the 'Remove notifications' button: https://q3k.org/u/67e92320ba45ca8c58d4945b2afb8255ce869f64f0e6f941a7366d17a0bee438.png

Not a bug, but a comment (maybe a feature suggestion...)
I often see people "add links" to a song and it comes through broken, like this:
https://soundcloud.com/https://soundcloud.com/blisshenrikjose/bliss-fighting-flyboy-deluxe
It's because we only expect part of the URL, such as the username and song title, or maybe just a numeric ID. So if you put the whole ID in, it adds the "template" to that, and produces nonsense like the above.
I see the point of this -- it prevents people adding random unrelated URLs -- but the error is quite common. We should either make it clearer what you have to do (instead of hovering over the help icon, which many users seem to miss), or we could perhaps validate what they type in the box. So if it's an ID, for example, we would reject anything non-numeric with an explanatory message.

Descriptions seems broken on this compo page https://scenestream.net/demovibes/compilation/258/

Tezar: yeah, unfortunately a lot of the compilation pages are broken, with the HTML markup being rendered as text. It's known about but not fixed yet. (Anyone know what the issue is? Could we fix it as simply as search-replace on some db fields that might have "&lt;" instead of "<" and so on?)
...haha I wonder what are the chances of this message not getting fked up in transit.

VodkaBot discovered that songs with "exotic" characters sometimes can't be requested (the green request button fades but nothing happens). This happened with Naive Love when the title contained Cyrillic characters: "Наивная любовь (Naive Love)". I renamed it to "Naive Love" and it became requestable.

A display bug: visit the Unknown artist Unknown and note that the song Commando intro [amstrad] appears twice in the list of songs. It's not a duplicate: it's the same identical song listed twice in a row.

It seems you can add the same link more than once. For example Showdown at Sunset has DZ two times, with same 82614 number.

Tilbury Fair has a duplicate external link. I don't know how common that is, first one I noticed.

Different Ways another one with duplicate external link.

There are (I think) *lots* of dupe external links and we should remove them in an automated way. I believe it was Nate who added a lot of these and I (and probably others) would check that they were correct and approve them, without realising the link might already be there. I think in some cases it's the same page but a different URL (like http and https, or with and without www).

Yes, these can be automatically removed. Additionally, the link-adding code can be modified to avoid inserting duplicate links in the first place -- I'm surprised there were no constraints on that as it seems like an obvious bug. Thanks for reporting!

Tezar: Regarding the gobbledygook on https://scenestream.net/demovibes/compilation/258/, this (and other instance) are apparently due to bad screen-scraping / parsing at the time we were reincarnating the site. I'm not sure the original info can be reconstructed but I'll look into it.