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

RE: [XaraXtreme-dev] Current/default font

Hi Martin,

Technically, it's correct to keep records as small as possible but that
becomes less important for records that are rarely used and in the end
if the information carried by a record is required then it's got to be
present no matter how big the record is. (Personally I'm not sure record
size is so much of an issue these days but we'd have to formally agree a
new position on that.)

I know it's frustrating trying to navigate through all the old code and
all the old decisions and assumptions behind them, many of them hidden
to us now.

I notice that Xtreme's in-memory tree has a font face attribute stored
locally when "Times New Roman" is applied to a Story. That explains why
the XAR file has the equivalent records. I'll try to find out why that
attribute is not being optimised out and let you know whether it's
deliberate or accidental...


> -----Original Message-----
> From: owner-dev@xxxxxxxxxxxxxxxx 
> [mailto:owner-dev@xxxxxxxxxxxxxxxx] On Behalf Of Martin Wuerthner
> Sent: 27 July 2006 14:06
> To: dev@xxxxxxxxxxxxxx
> Subject: RE: [XaraXtreme-dev] Current/default font
> Hi Phil,
> In message 
> <5056CBC646CB4047BB26120F4377DB71DC8425@xxxxxxxxxxxxxxxxxxx 
> >
>           "Phil Martin" <Phil@xxxxxxxx> wrote:
> > Yes I did mean "Times New Roman" and the presence of those 
> records in 
> > the XAR document does not necessarily indicate a bug: The 
> text system 
> > could be writing out font face records deliberately even when font 
> > face attributes are not present in the in-memory tree.
> In theory, it could, but I do not think it does (do a grep 
> for TAG_FONT_TYPEFACE and you see that only the attribute 
> node saving code itself writes this record), and if it did, 
> it would still be wrong. 
> When I designed the file records for the new attributes I was 
> gently told off for using a few unnecessary *bytes* in my 
> first design and was asked to make sure the file is as small 
> as possible. Surely, there would be absolutely no point in 
> the text system writing superfluous
> *records* to the file, would there?
> Martin