ARTICLE - Why the MS Office file formats is so complicated

[Alex Hewitt]

On Wed, 2008-02-20 at 17:23 -0500, Ben Scott wrote:
> On Wed, Feb 20, 2008 at 3:58 PM, Alex Hewitt <hewitt_tech at comcast.net> wrote:
> >  A friend of ours wrote a bunch of recipe files using something called
> >  Microsoft Write.
> 
>   Yah, "Windows Write" is/was one of the "accessories" that came with
> Windows 3.x.  It morphed into "WordPad" in Windows 95 and later.
> WordPad still exists.  It won't write the Write (hah) format anymore,
> but it can read it, and save in some variant of the RTF format.
> 
> >  Theoretically Microsoft Word is supposed to be able to read such files
> >  but I found that the version I was using (Word 2003) wouldn't.
> 
>   Curious.  My install of Word 2003 can.  Are you sure you installed
> all the import/export filters?  If you did a "Minimal" or "Custom"
> install (instead of the mondo-huge "Full"), I don't think those are
> all included by default.

I sit corrected! ;^) Word 2003 complains about the file saying in effect
"this might be a virus but I have a converter that will convert it" and
it does. I think the original reason I wrote the filter was because our
friend didn't have Word and I didn't want to manually edit her 83 files.
I'll see if Word can be called from the command line to do the
converting.

-Alex

> 
> > Writing a filter in Python was trivial and I was  able to convert the
> > files to plain text.
> 
>   For future reference, the strings(1) command can be used to much the
> same effect.
> 
> > ... the file itself in ASCII, a series of bytes  again in non-ASCII,
> > followed by a repeat of some of the original ASCII.
> 
>   That sounds very similar to the MS Word .DOC format, and I bet
> they're related.  DOC files do not interleave the formatting with the
> text, as (for example) HTML or Word Perfect did.  Instead, all the
> plain text is stored in one blob, and then the formatting information
> is stored in a different blob.  The formatting directives have
> "pointers" to the position of the text they effect.
> 
>   The "repeat" you describe is not actually a repeat, but a follow-on
> save.  Word and friends work in an interesting fashion.  You open the
> file, and it loads the base text blob described above.  You start
> making your changes.  Those changes go into an undo buffer.  That undo
> buffer is actually backstored on the disk in temporary files.
> (That's why a directory containing Word files people are busy editing
> accumulates lots of odd temp files until they close the original.)
> 
>   When you invoke "Save", the undo buffer -- essentially like a "diff"
> -- gets tacked on to the end of the main file.  This made saves fast
> on slow computers already overburdened by Microsoft bloatware.  Loads
> were slower, of course, but the reasoning was that people care about
> save speed more than load speed.    As you can imagine, if there are
> lots of saves, rebuilding the text is not so easy as running
> strings(1) on it.
> 
>   In Word, if you turn off "Fast Saves", it writes out a full, unified
> version of the text instead.  This became the default at some point --
> I have no idea when.
> 
> > But the interesting thing was that I  couldn't easily find a Microsoft tool that
> > understood the format which originated with Windows 95 or an earlier version
> > of Windows.
> 
>   Start -> Programs -> Accessories -> WordPad
> 
>   My copy of Win XP Pro opens .WRI files automatically in WordPad.  I
> just double-click the file.
> 
>   WordPad is an optional component for Windows.  Perhaps the computer
> was installed with a "minimalist" attitude, so various optional tools
> were not there when you needed them?
> 
> -- Ben
> _______________________________________________
> gnhlug-discuss mailing list
> gnhlug-discuss at mail.gnhlug.org
> http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/