Towards the ccid-3 and ccid-4 integration

Towards the ccid-3 and ccid-4 integration

[@ 2007-09-24 18:15 UTC (permalink / raw)]

Hi list,
  I'm currently trying to define the steps in order to get a good
patchset for the initial ccid-4 implementation. Thus, I'd like to
discuss with all of you some topics and procedures:

- Rewrite all the patches (from me and Tommi) and resubmit to Gerrit
as a unique patchset, as suggested by Ian. Besides, ask Gerrit to
overwrite the current ccid4 branch;
- Identify and provide common code between ccid-4 and ccid-3.
  1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
the ccid-4.c, as suggested by Ian and Arnaldo
  2 - for more complex functions, write a dccp_ccid3_lib.ko, as
suggested by arnaldo

Please, let me know if I've forgot something.

Leandro.

Re: Towards the ccid-3 and ccid-4 integration

[@ 2007-09-24 21:35 UTC (permalink / raw)]

2007/9/24, ツ Leandro Sales <leandroal@gmail.com>:
> Hi list,
>   I'm currently trying to define the steps in order to get a good
> patchset for the initial ccid-4 implementation. Thus, I'd like to
> discuss with all of you some topics and procedures:
>
> - Rewrite all the patches (from me and Tommi) and resubmit to Gerrit
> as a unique patchset, as suggested by Ian. Besides, ask Gerrit to
> overwrite the current ccid4 branch;
> - Identify and provide common code between ccid-4 and ccid-3.
>   1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
> the ccid-4.c, as suggested by Ian and Arnaldo
>   2 - for more complex functions, write a dccp_ccid3_lib.ko, as
> suggested by arnaldo
>
> Please, let me know if I've forgot something.
>
> Leandro.
>

Is there someone that would like to comment the proceeds that me and
Tommi will adopt for the initial ccid-4. Tommi, anything more?

Leandro.

Re: Towards the ccid-3 and ccid-4 integration

[Ian McDonald]

On 9/25/07, ツ Leandro Sales <leandroal@gmail.com> wrote:
> Hi list,
>   I'm currently trying to define the steps in order to get a good
> patchset for the initial ccid-4 implementation. Thus, I'd like to
> discuss with all of you some topics and procedures:
>
> - Rewrite all the patches (from me and Tommi) and resubmit to Gerrit
> as a unique patchset, as suggested by Ian. Besides, ask Gerrit to
> overwrite the current ccid4 branch;
> - Identify and provide common code between ccid-4 and ccid-3.
>   1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
> the ccid-4.c, as suggested by Ian and Arnaldo
>   2 - for more complex functions, write a dccp_ccid3_lib.ko, as
> suggested by arnaldo
>
> Please, let me know if I've forgot something.
>
> Leandro.

Sounds great to me.

Ian
-- 
Web1: http://wand.net.nz/~iam4/
Web2: http://www.jandi.co.nz
Blog: http://iansblog.jandi.co.nz

Re: Towards the ccid-3 and ccid-4 integration

[Gerrit Renker]

Sorry for the delay there were local bank holidays here.

|  - Rewrite all the patches (from me and Tommi) and resubmit to Gerrit
|  as a unique patchset, as suggested by Ian. Besides, ask Gerrit to
|  overwrite the current ccid4 branch;
|  - Identify and provide common code between ccid-4 and ccid-3.
|    1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
|  the ccid-4.c, as suggested by Ian and Arnaldo
|    2 - for more complex functions, write a dccp_ccid3_lib.ko, as
|  suggested by arnaldo
The ideas sound good, and I think it is a good idea to factor out what is
not needed - just unclear which patch set to use, the latest so far is from
Thursday?

Re: Towards the ccid-3 and ccid-4 integration

[Tommi Saviranta]

On Mon, Sep 24, 2007 at 15:15:57 -0300, ツ Leandro Sales wrote:
> - Identify and provide common code between ccid-4 and ccid-3.

I'm probably little too much of a need-get-my-hands-dirty kind of guy,
but I think this part usually works out the best while actually doing
the split. I don't want to scare people away by telling this, but if I
was working on the split alone, I'd start by moving everything in
ccid3.c to lib/ccid3.c, and then moving stuff back to ccid3.c in pieces
as certain functions would have to be rewritten for CCID 4.

>   1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
> the ccid-4.c, as suggested by Ian and Arnaldo
>   2 - for more complex functions, write a dccp_ccid3_lib.ko, as
> suggested by arnaldo

There is very little more to say about this, really. The code should and
needs to be shared. I just have this bad feeling that ccid3.c will
become nothing but a skeleton file that does all the work in
dccp_ccid3_lib.

Re: Towards the ccid-3 and ccid-4 integration

[Ian McDonald]

On 9/27/07, Tommi Saviranta <wnd@iki.fi> wrote:
> On Mon, Sep 24, 2007 at 15:15:57 -0300, ツ Leandro Sales wrote:
> > - Identify and provide common code between ccid-4 and ccid-3.
>
> I'm probably little too much of a need-get-my-hands-dirty kind of guy,
> but I think this part usually works out the best while actually doing
> the split. I don't want to scare people away by telling this, but if I
> was working on the split alone, I'd start by moving everything in
> ccid3.c to lib/ccid3.c, and then moving stuff back to ccid3.c in pieces
> as certain functions would have to be rewritten for CCID 4.
>
Yes it's best to experiment and then tidy later often.

> >   1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
> > the ccid-4.c, as suggested by Ian and Arnaldo
> >   2 - for more complex functions, write a dccp_ccid3_lib.ko, as
> > suggested by arnaldo
>
> There is very little more to say about this, really. The code should and
> needs to be shared. I just have this bad feeling that ccid3.c will
> become nothing but a skeleton file that does all the work in
> dccp_ccid3_lib.
>
And that's fine. There will be some methods that are different - like
calculating the ipi - and this would then be in ccid3.c and ccid4.c
but not in dccp_ccid3_lib.c

Ian
-- 
Web1: http://wand.net.nz/~iam4/
Web2: http://www.jandi.co.nz
Blog: http://iansblog.jandi.co.nz

Re: Towards the ccid-3 and ccid-4 integration

[@ 2007-09-26 23:51 UTC (permalink / raw)]

2007/9/25, Gerrit Renker <gerrit@erg.abdn.ac.uk>:
> Sorry for the delay there were local bank holidays here.
>
> |  - Rewrite all the patches (from me and Tommi) and resubmit to Gerrit
> |  as a unique patchset, as suggested by Ian. Besides, ask Gerrit to
> |  overwrite the current ccid4 branch;
> |  - Identify and provide common code between ccid-4 and ccid-3.
> |   1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
> |  the ccid-4.c, as suggested by Ian and Arnaldo
> |   2 - for more complex functions, write a dccp_ccid3_lib.ko, as
> |  suggested by arnaldo
> The ideas sound good, and I think it is a good idea to factor out what is
> not needed - just unclear which patch set to use, the latest so far is from
> Thursday?
> -
> To unsubscribe from this list: send the line "unsubscribe dccp" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Hi,
  Gerrit, the latest so far is from Friday, the mine, which it was
similar to the Tommi patch, so then I just patches the differences
between the mine and the tommi's patchset. But I'm put all the patches
together (the mine and tommi as one patchset and I will ask you to
overwrite the current ccid4 branch. Besides, I'm going through the
ideas of dccp_ccid3_lib.ko.

Leandro.

Re: Towards the ccid-3 and ccid-4 integration

[Gerrit Renker]

Quoting Tommi Saviranta:
|  >   1 - minor functions from ccid-3.c to ccid-3.h and include it (.h) in
|  > the ccid-4.c, as suggested by Ian and Arnaldo
|  >   2 - for more complex functions, write a dccp_ccid3_lib.ko, as
|  > suggested by arnaldo
|  
|  There is very little more to say about this, really. The code should and
|  needs to be shared. I just have this bad feeling that ccid3.c will
|  become nothing but a skeleton file that does all the work in
|  dccp_ccid3_lib.
|  
I have thought about this - and had an idea to eliminate one extra library:

 * use a .ko file for the functions called by both CCID3 and CCID4 as suggested by Arnaldo
 * but there is already dccp_tfrc_lib which hosts a major number of functions used by CCID3/4
 * hence I think that you could proceed by
	- add a CCID-3 specific object file to dccp_tfrc_lib-y in net/dccp/ccids/lib/Makefile
          (eg. ccid3_funcs.o or whatever)
	- the file could take the shared functions, ccid3.c could then be stripped to smaller functions
	  which wrap around the library functions
	- tfrc.h could take the declarations
 * this way one library less to look after.

Ian/Arnaldo may also have ideas or opinions on this.

Re: Towards the ccid-3 and ccid-4 integration

[Ian McDonald]

On 9/27/07, Gerrit Renker <gerrit@erg.abdn.ac.uk> wrote:
>  * but there is already dccp_tfrc_lib which hosts a major number of functions used by CCID3/4
>  * hence I think that you could proceed by
>         - add a CCID-3 specific object file to dccp_tfrc_lib-y in net/dccp/ccids/lib/Makefile
>           (eg. ccid3_funcs.o or whatever)
>         - the file could take the shared functions, ccid3.c could then be stripped to smaller functions
>           which wrap around the library functions
>         - tfrc.h could take the declarations
>  * this way one library less to look after.
>
I agree. I came up with this idea too and then I promptly forgot it!

Ian
-- 
Web1: http://wand.net.nz/~iam4/
Web2: http://www.jandi.co.nz
Blog: http://iansblog.jandi.co.nz