Re: PATCH: cdrecord: avoiding scsi device numbering for ide devices

[Martin Schlemmer]

On Tue, 2004-08-10 at 15:03, Joerg Schilling wrote:
> >From: David Woodhouse <dwmw2@xxxxxxxxxxxxx>
> 
> >On Tue, 2004-08-10 at 14:47 +0200, Joerg Schilling wrote:
> >> Cdrecord does not read /etc/cdrecord.conf
> 
> >And the world is flat.
> 
> >shinybook /home/dwmw2 $ strace -e open cdrecord -inq 2>&1 | grep /etc/cdrecord.conf
> >open("/etc/cdrecord.conf", O_RDONLY)    = 3
> >open("/etc/cdrecord.conf", O_RDONLY)    = 3
> >open("/etc/cdrecord.conf", O_RDONLY)    = 3
> >open("/etc/cdrecord.conf", O_RDONLY)    = 3
> 
> It seems that you like to constantly discredit yourself :-(
> 
> What you use if DEFINITELY NOT cdrecord.
> 

While I would rather have said something like:

  What you use is DEFINITELY a PATCHED cdrecord

it is not the original:

---
# strace -e open cdrecord -inq 2>&1 | grep '/etc/.*cdrecord'
open("/etc/default/cdrecord", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/etc/default/cdrecord", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/etc/default/cdrecord", O_RDONLY) = -1 ENOENT (No such file or directory)
---

stupid question: why this name and placement of the config file?
I would think that /etc/cdrecord.conf or /etc/cdrecord/defaults might
have made more sence, but this is out of a mostly linux pov.

Also, out of a user point of view:

1)  What is the big issue with supporting -dev=/dev/xxx ?  As I did
    point out above, I have limited experience with solaris and fbsd,
    but don't they all support and use /dev/ to access devices ?  It
    really is not a new way to access devices in linux as you tried
    to point out earlier - and it should be possible to implement it
    on all *nix OS's ?  Sure, maybe the exact kernel interface might
    change, but your device node should always (mostly?) point you
    in the correct direction if you know the correct one to use?
    I and I am sure many (most?) other users that have ide/usb devices
    do not want to worry about scsi topology, etc - that and scsi
    emulation adds extra overhead that we with older machines do not
    want.

    Also, it would be less confusing - on 2.6 currently -scanbus do
    not even detect any cdrw devices (sure, not a problem of yours).
    But I still remember the first time I tried to use cdrecord - and
    later with 2.5 and the -dev=x,y,z syntax ... I just went with
    gtoaster and did not look back as cli writing is not an issue
    here.

2)  As pointed out, some error messages could use some tweaking.
    There was also some nice examples ...


I hope this was put nicely enough (although a tad clueless), and valid
a similar response.  If not, please refrain from answering.


Regards,

-- 
Martin Schlemmer

Attachment: signature.asc
Description: This is a digitally signed message part