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

Re: DDP CVS commit by jseidel: ddp/manuals.sgml/release-notes/de release-note ...



On Sun, Jan 09, 2005 at 08:04:58PM +0100, Geert Stappers wrote:
> On Sun, Jan 09, 2005 at 06:24:17PM +0100, Jens Seidel wrote:
> > On Sun, Jan 09, 2005 at 05:57:22PM +0100, Geert Stappers wrote:
> > > On Sun, Jan 09, 2005 at 09:30:55AM -0700, DDP CVS wrote:
>  <snip/>
> > > > Modified files:
> > > > 	manuals.sgml/release-notes/de: release-notes.de.sgml 
> > > > 
> > > > Log message:
> > > > 	use latin1 encoding instead of HTML entities to simplify proofreading
> > > >       and to increase compatibility with various tools
> > > 
> > > I think you broke compatibility with XML & SGML tools. [1]
> > > [1] both are ASCII only.
> > 
> > The old code using &uuml; and &szlig; in each third word makes the SGML
> > source code nearly unreadable. Please note that the previous version
> > already contained a mixture of latin1 and ASCII.
[snip]
> > I also found a missing umlaut in the output (last PDF page) because
> > <url id="&url-debian-i18n;" name="&Uuml;bersetzungen">
> > was used but which works great using name="Übersetzungen".
> 
> My concern is that the source is not pure 7-bit ASCII.

Please note that all versions of Release Notes (except English one)
use 8bit characters. Do you really expect that a Japanese translator
writes &entity1;&entity2;&entity3;...?

> It should ASCII only for XML and SGML.

But debiandoc-sgml supports all common locales, especially latin1,
latin2,...

> Jens, that you spend time on the release notes is good.
> I do respect that.
> 
> But your arguments to break stuff are poor.

Maybe. I agree that pure ASCII has advantages but the current file can
easily converted to another locale using iconv, recode, konwert, ...

> * it is hard to proofread gr&ouml;&szlig;e dateien

It is!

> Consider the SGML source as computer program source
> and proofreading is running the programm.
> Each "bug" you find, has to be modified in the source.
> The edit-compile-test cycle can indeed be boring,
> you shouldn't cheat by implementing "compiled blobs"
> in the source code.

I would really like to know other people's opinions.

> As one volunteer to another volunteer:
> 
>   Please revert the latin1 changes

First I will wait for more feedback. If the majority agrees with you I
will revert my changes (but what about name="&Uuml;bersetzungen" which
doesn't work for url tags?).

Jens



Reply to: