cytune: misnamed long options

cytune: misnamed long options

[Benno Schulenberg]

Hi,

According to the code and the man page of cytune, the option -g
shows the current values of the threshold and the flush timeout,
and option -G shows the default values.  But the corresponding
long options of those two are: --get-threshold and --get-flush.
Confusing.  Shouldn't those be changed to --get-current and
--get-defaults?  (Or the first one maybe just --get?)  However,
the docs say not to change any options.  What to do?  Sami?

Regards,

Benno

-- 
http://www.fastmail.fm - The professional email service

Re: cytune: misnamed long options

[Sami Kerola]

On 30 April 2014 09:52, Benno Schulenberg <bensberg@justemail.net> wrote:
> According to the code and the man page of cytune, the option -g
> shows the current values of the threshold and the flush timeout,
> and option -G shows the default values.  But the corresponding
> long options of those two are: --get-threshold and --get-flush.
> Confusing.  Shouldn't those be changed to --get-current and
> --get-defaults?  (Or the first one maybe just --get?)  However,
> the docs say not to change any options.  What to do?  Sami?

Hi Benno,

The recommendation never to change options tries to protect a Random J
Developer in breaking existing scripts. If options are deprecated at
least there should be a backwards compatibility with a deprecation
warning for some time (couple years).

Looking the kernel interface

http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/tty/cyclades.c#n2639

the 'threshold' keyword seems to be ok. Use of word 'flush' is a bit
odd, and 'timeout' would be closer match with what the kernel is
doing. Karel, are you OK with s/flush/timeout/ ABI change?

After looking sys-utils/cytune.c for a moment I feel the command could
be improved various way. I could aim to do that before v2.25, unless
someone else wants to do more or less similar things what was done
with kill(1).

-- 
Sami Kerola
http://www.iki.fi/kerolasa/

Re: cytune: misnamed long options

[Karel Zak]

On Wed, Apr 30, 2014 at 12:25:49PM +0100, Sami Kerola wrote:
> On 30 April 2014 09:52, Benno Schulenberg <bensberg@justemail.net> wrote:
> > According to the code and the man page of cytune, the option -g
> > shows the current values of the threshold and the flush timeout,
> > and option -G shows the default values.  But the corresponding
> > long options of those two are: --get-threshold and --get-flush.
> > Confusing.  Shouldn't those be changed to --get-current and
> > --get-defaults?  (Or the first one maybe just --get?)  However,
> > the docs say not to change any options.  What to do?  Sami?
> 
> Hi Benno,
> 
> The recommendation never to change options tries to protect a Random J
> Developer in breaking existing scripts. If options are deprecated at
> least there should be a backwards compatibility with a deprecation
> warning for some time (couple years).
> 
> Looking the kernel interface
> 
> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/tty/cyclades.c#n2639
> 
> the 'threshold' keyword seems to be ok. Use of word 'flush' is a bit
> odd, and 'timeout' would be closer match with what the kernel is
> doing. Karel, are you OK with s/flush/timeout/ ABI change?

 NACK, it seems better to be backwardly compatible than perfect.

    Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com

Re: cytune: misnamed long options

[Sami Kerola]

On 6 May 2014 10:26, Karel Zak <kzak@redhat.com> wrote:
> On Wed, Apr 30, 2014 at 12:25:49PM +0100, Sami Kerola wrote:
>> On 30 April 2014 09:52, Benno Schulenberg <bensberg@justemail.net> wrote:
>> > According to the code and the man page of cytune, the option -g
>> > shows the current values of the threshold and the flush timeout,
>> > and option -G shows the default values.  But the corresponding
>> > long options of those two are: --get-threshold and --get-flush.
>> > Confusing.  Shouldn't those be changed to --get-current and
>> > --get-defaults?  (Or the first one maybe just --get?)  However,
>> > the docs say not to change any options.  What to do?  Sami?
>>
>> Hi Benno,
>>
>> The recommendation never to change options tries to protect a Random J
>> Developer in breaking existing scripts. If options are deprecated at
>> least there should be a backwards compatibility with a deprecation
>> warning for some time (couple years).
>>
>> Looking the kernel interface
>>
>> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/tty/cyclades.c#n2639
>>
>> the 'threshold' keyword seems to be ok. Use of word 'flush' is a bit
>> odd, and 'timeout' would be closer match with what the kernel is
>> doing. Karel, are you OK with s/flush/timeout/ ABI change?
>
>  NACK, it seems better to be backwardly compatible than perfect.

The NACK seems to indicate the change

https://github.com/kerolasa/lelux-utiliteetit/commit/2453ab9f64de49bb5a1b9a0bcd3653cd761a22b2

should not warnx(), nor be marked deprecated. Or to say other way, the
old long options should be silently accepted until end of universe.
Let me fix that (later the day when I'm not in office).

-- 
Sami Kerola
http://www.iki.fi/kerolasa/