gnustep-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Progress of text system


From: Alexander Malmberg
Subject: Re: Progress of text system
Date: Mon, 24 Feb 2003 00:55:16 +0100

Richard Frith-Macdonald wrote:
[snip]
> However, from Alexanders description, the current layout contravenes
> the standard usage of classes/class clusters ...
> NSLayoutManager should be the abstract class providing the general
> implementation, and GSLayoutManager (or perhaps a more meaningful name
> like GSWesternLayoutManager) ought to be the subclass providing the
> specialised/optimised implementation.   When NSLayoutManager is
> instantiated, it would provide a concrete instance of GSLayoutManager
> to actually do the work.  This is the way the naming conventions for
> class clusters normally work, and doing things the other way round
> might confuse programmers.  However, this is just a matter of the
> naming conventions for the classes and does not effect the real
> structure.

The naming is a bit unfortunate, but it's necessary for OPENSTEP
compatibility, and it's a fairly small price to pay for that. If
NSLayoutManager was the non-layout-specific version, subclasses of
NSLayoutManager would not work as on OS, and there would be no way of
creating plain GSLayoutManager instances.

> So NSLayout manager would provide the capabilities of storing layout
> runs and glyph storage etc, wheras the concrete subclass would provide
> the optimised generation algorithms.  Indeed, if the concrete
> subclasses provided no ivars of their own, it would be possible for
> them to mutate at runtime depending on the
> languages used ... so when NSLayoutManager is instantiated it could
> create a  GSWesternLayoutManager, but storage of chinese text might
> make it mutate into a slower but more general GSChineseLayoutManager
> for instance.

So far, there is nothing that would prevent mutating between
NSLayoutManager and other (hypothetical) text-view-supporting layout
managers.

[mail #2]
Richard Frith-Macdonald wrote:
> PS.
> Please note ... criticism of your changes to the text system are not 
> meant to be unappreciative ... if it sounds like I'm saying that you've 
> failed  to address some of the issues (mainly the character<=>glyph 
> relationships) that is not to imply that that I don't understand/like 
> the improvements you have made.  After all, Fred, Nicola and I have all 
> failed to do anything much about this in the last year .... this stuff 
> is not simple/easy!

True, the text system is a big and complicated thing, and it took a lot
of time to work through all the parts and see all the issues. I do feel,
however, that the structure to handle everything is in place, so
although there are things left to implement, I now know where to place
them and how they're supposed to fit in. :-)

(Or why they don't fit in; the text system isn't meant to handle
everything, and there are trade-offs involved.)

- Alexander Malmberg




reply via email to

[Prev in Thread] Current Thread [Next in Thread]