[Abcde-users] Has a "ranking" for abcde.conf settings ever been considered?

[Gordon Findlay]

Just to close this off from my POV anyway:


1. I have installed the patch Gabriel circulated, and tested it with 4 CDs,
and various combinations for CDDBMETHOD, and it all seems to have worked.

2. Using the modified permissions (i.e. running SUID root) on cdda2wav did
give me CD-TEXT as expected when CDDB lookup failed.

3. I tried installing icedax in place of cdda2wav on openSUSE.  It installs
with normal permissions for executables - no suid - and runs fine for
normal users.

Thanks Gabriel

Slainte
Gordon

On Fri, Sep 2, 2016 at 10:09 PM, Gabriel Rosenkoetter <gr at eclipsed.net>
 wrote:

> On 26 Aug, 2016, at 18:58EDT, Gabriel Rosenkoetter <gr at eclipsed.net>
> wrote:
> > I’m thinking in particular of CDDBMETHOD here, but there are probably
> other places where this might be handy.
>
> Short version:
>
> The attached patch (see also https://abcde.einval.com/
> bugzilla/show_bug.cgi?id=42), if applied to your local abcde, will make
> CDDBMETHOD accept a comma-separated list of CDDB resolution methods. If
> you’ve only got one thing set for that in your CDDBMETHOD (that is, you
> have no commas there), it shouldn’t break anything. I would recommend
> setting “CDDBMETHOD=musicbrainz,cddb”, but that’s just me.
>
>
On Sat, Sep 3, 2016 at 6:05 PM, Gordon Findlay <gfindlay.linux at gmail.com>
wrote:

> As Gabriel commented, abcde did not read any CD-TEXT information for me.
> These details are what I have so far:
>
> 1: abcde uses icedax or cdda2wav (in do_cdtext() ) to read it. I am using
> openSUSE Leap 42.1, in which icedax is not installed, so abcde tries to use
> cdd2a2wav. That *is* installed (version 3.01).
>
> 2: in a default installation of openSUSE, a standard user cannot run
> cdda2wav, due to privilege issues like these:
>
> cdda2wav: Operation not permitted. Cannot regain process capabilities.
> cdda2wav: Insufficient 'file read' privileges. You will not be able to
> open all needed devices.
> cdda2wav: Insufficient 'file write' privileges. You will not be able to
> open all needed devices.
> cdda2wav: Insufficient 'device' privileges. You may not be able to send
> all needed SCSI commands, this my cause various unexplainable problems.
>
> cdda2wav: Insufficient 'priocntl' privileges. You may get jitter.
> cdda2wav: Insufficient 'network' privileges. You will not be able to do
> remote SCSI.
> cdda2wav: Operation not permitted. Cannot regain process capabilities.
>
> cdda2wav: Permission denied. Cannot open '/dev/sg1'. Cannot open or use
> SCSI driver.
> cdda2wav: For possible targets try 'cdda2wav -scanbus'. Make sure you are
> root.
>
> cdda2wav works fine for root, or using sudo
>
> 3: The default (at least for SUSE) perms are
> -rwxr-xr-x 1 root root 262944 Oct 25  2015 cdda2wav
>
> chmod u+s cdda2wav resolves the problem, so normal users can execute
> cdda2wav. This is the quick fix.
>
> 4. Running programs suid root like this is anathema to some. According to
> the cdrtools web site it is possible to be more subtle using the setcap to
> assign more fine-grained privileges. A default installation of openSUSE
> doesn't include setcap either :-)
> ​
> I will test the patch for a ranking list of CDDBMETHOD in a little while -
> tomorrow I hope.
>
> Thanks to Gabriel for forcing me to investigate this. I wish I had checked
> what was going on before I used a different technique altogether to get CDs
> with no CDDB/MusicBrainz entries (I used cd_info, a dodgy script, and a lot
> of cut and paste). cd_info gives different formatted output to either
> icedax or cdda2wav. so I guess using it as CDTEXT_READER in abcde would be
> difficult.
>
> Slainte
> Gordon
>
>
> On Fri, Sep 2, 2016 at 10:09 PM, Gabriel Rosenkoetter <gr at eclipsed.net>
> wrote:
>
>> On 26 Aug, 2016, at 18:58EDT, Gabriel Rosenkoetter <gr at eclipsed.net>
>> wrote:
>> > I’m thinking in particular of CDDBMETHOD here, but there are probably
>> other places where this might be handy.
>>
>> Short version:
>>
>> The attached patch (see also https://abcde.einval.com/bugzi
>> lla/show_bug.cgi?id=42), if applied to your local abcde, will make
>> CDDBMETHOD accept a comma-separated list of CDDB resolution methods. If
>> you’ve only got one thing set for that in your CDDBMETHOD (that is, you
>> have no commas there), it shouldn’t break anything. I would recommend
>> setting “CDDBMETHOD=musicbrainz,cddb”, but that’s just me.
>>
>> Please let me know if you apply this patch and have problems as a result.
>> I don’t think i broke anything, but it’s really hard to test for unknowns.
>>
>> Long version:
>>
>> I’ve cc’ed a few people explicitly here (and I would encourage you to
>> drop the explicit ccs if you’re responding to this to the list, in order to
>> reduce everybody’s inbox noise):
>> * Steve, because he and I have had some off-list exchanges about this,
>> and … I guess I just want to make his inbox catchup worse? Huh, maybe I
>> shouldn’t actually send you two copies of this. (And now I’m not.)
>> * Gordon, because he responded, inadvertently dropping the list address
>> (we discussed that, and agreed that I’d go ahead and summarize our
>> conversation back here) to my original email about this (quoted in
>> abbreviation above) saying, broadly, “Yeah, i want that too!”. Gordon also
>> made some good points about how, in absence of MusicBrainz or CDDB records
>> for a given CD, maybe just fall back to CD-TEXT? The existing code does
>> fall through to CD-TEXT in something like that case, but I’m not convinced
>> it works the way a theoretical new user thinks it should. The details of
>> this get into CDTEXT_READER behavior, and that’s just not a thing I’ve
>> looked at carefully. Gordon, please feel to expand on this.
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.einval.com/pipermail/abcde-users/attachments/20160904/518345ff/attachment.html>