--===============2019884586==
Content-Type: multipart/signed;
  boundary="nextPart1319701.dIyN2sTBqu";
  protocol="application/pgp-signature";
  micalg=pgp-sha1
Content-Transfer-Encoding: 7bit

--nextPart1319701.dIyN2sTBqu
Content-Type: text/plain;
  charset="euc-jp"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

On Sunday 26 September 2004 22:23, Cyrille Berger wrote:
> I was about to commit a change :
>
> Change in the KisFilter API :
> * the process now make a distinction between the source paint device and
> the destination
>
> Which mean we have two functions in filter :
>  virtual void process(KisPaintDeviceSP src, KisPaintDeviceSP dst,
> KisFilterConfiguration*, const QRect&, KisTileCommand* ) =3D 0;
>  void process(KisPaintDeviceSP, KisFilterConfiguration*, const QRect&,
> KisTileCommand* );
>
> Does it allow to do what you want ?

I think it should help Bart in that it gives a cleaner, but doesn't this st=
ill=20
mean that you need to create a temporary target paint device? That's the bi=
g=20
problem. Of course, if we could invent something like a 'lazy' paint device=
=20
that only allocates tiles if and when and where pixels ares changed, that=20
should help enormously. That's more or less what Casper was doing with the=
=20
autoextending layers, isn't it?

=2D-=20
Boudewijn Rempt | "Geef mij maar zuurtjes."
http://www.valdyas.org/fading/index.cgi

--nextPart1319701.dIyN2sTBqu
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQBBV7ekdaCcgCmN5d8RAqD9AKDOVx0r+BnlMM0Kk6ztd4UF9NwuewCgmh2x
givGNbW1jIScLZYWx+yZ9Lk=
=+kQD
-----END PGP SIGNATURE-----

--nextPart1319701.dIyN2sTBqu--


--===============2019884586==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
kimageshop mailing list
kimageshop@kde.org
https://mail.kde.org/mailman/listinfo/kimageshop

--===============2019884586==--