mt-daapd 1005, song lenght

[fizze]

Initially I thought it to be a roku error, but they claim to get song lengths from the server.

I recently encoded some VBR’s with high bitrate, and the lenght that mt-daapd reported was 40:32, as opposed to a real length of about 4:30.

How can I be of service to help fixing ? 😉

[rpedde]

First, try scanning it with “scan_type” set to 2. That does the “slow but accurate” scan. If it shows up okay, then it’s because the file was tagged with a file that didn’t add an xing vbr tag. Bad tagger!

If it still shows up wrong, then email it to me at ron at pedde.com, and I can fix it.

— Ron

[fizze]

Ah, alright. Scan type is currently set to 0 :rolleyes:
I’ll play with that.

[fizze]

Well, I set it to 2 and my slug stopped responding.
I even gave it a whole day to scan, but I eventuelly had to reboot.

So, are there any tag-fixing tools so I could adjust those files on / from my PC?

[rpedde]

@fizze wrote:

Well, I set it to 2 and my slug stopped responding.
I even gave it a whole day to scan, but I eventuelly had to reboot.

So, are there any tag-fixing tools so I could adjust those files on / from my PC?

You can do it from iTunes, by telling it to remove tags.

You could also try this http://www.mp3tag.de/en/index.html.

But I’d like to see the bad files. You could email it to me at ron at pedde.com.

— Ron

[fizze]

Ok, I’ll give it a shot once Im home.
In fact this looks nice, as it can strip/convert case of tags as well.

Wouldnt it be kinda cool to have the mt-daapd DB be case insensitive ?
wouldnt be to hard to add a few UPPER() or LOWER()’s in the queries, would it ?

edit:
I just took a peek at db-sql.c, its where the browse magic happens, right ?
Being an Oracle guy I donno much about the features sqlite3, or the other backends, but I think those queries could be sped up.
Things like the double selects to get the count can be acheived by %%rowcount or cols, given sqlite3 supports that.

Too bad i dont have a running cross-compiler environment for my slug so I could play with it. But I still can perform and test the queries manually on the slug.

[rpedde]

@fizze wrote:

Ok, I’ll give it a shot once Im home.
In fact this looks nice, as it can strip/convert case of tags as well.

Wouldnt it be kinda cool to have the mt-daapd DB be case insensitive ?
wouldnt be to hard to add a few UPPER() or LOWER()’s in the queries, would it ?

I could go either way on that. Like queries on mysql are already insensitive, so if everything was, I wouldn’t argue with that.

I just took a peek at db-sql.c, its where the browse magic happens, right ?
Being an Oracle guy I donno much about the features sqlite3, or the other backends, but I think those queries could be sped up.
Things like the double selects to get the count can be acheived by %%rowcount or cols, given sqlite3 supports that.

Yeah, but in the case of an indexed search I need not the rows returned, but the total rows without the offset/limit.

Also, it has to work with sqlite, mysql, and postgres at a minimum, which means sql-92 only.

If you can come up with performance enhancements that meet that criteria, I would certainly welcome them. I’m not much of a sql guy. Enough to hack about a bit, but by no means a dba.

As far as the slug goes, I want to add back the gdbm backend before a 1.0 release — it’s far snappier than sqlite on the slug.

— Ron

[Author]

Posts