[LSC] LDraw File Format Specification - Timetable

[William Howard]

All,

The current file format spec is here - http://www.ldraw.org/Article45.html 
The working copy of the new spec is here - http://www.ldraw.org/Article408.html
(The old draft spec is here - http://www.ldraw.org/Article218.html)

I've tried to add some text about the real-world approximations, but these are only my views so may not be shared by all.

I've also added comments/reasons to the other areas (use of colour 24 for line type 1 and polygons, and abutting polygons)

I appreciate that August is a terrible month in which to get anything done due to holidays, but I think we should aim to wrap these last few issues up by 1st Sept.  That way we can post the request for comments on Lugnet, give the community two weeks to respond, deal with any comments in the last two weeks of Sept and ratify the spec by 1st Oct - when we have to publish a summary of the years work.  (we can probably just update this page as our summary report - http://www.ldraw.org/Article399.html)

W
  ----- Original Message ----- 
  From: William Howard 
  To: LDraw Standards Committee 
  Sent: Monday, July 16, 2007 10:47 AM
  Subject: Re: [LSC] LDraw File Format Specification - Comments please


  All,

  I've gone back over all the emails and tried to edit the consensus views into the draft spec.

  There is one area that still needs to be discussed/agreed - that of the "real world approximations"

  There are still a few areas where the reasoning needs explaining and I'll try to expand these areas in the next few days.

  If I've missed anything, mis-represented anything, or just got anything plain wrong, please respond here.

  William
    ----- Original Message ----- 
    From: Travis Cobbs 
    To: LDraw Standards Committee 
    Sent: Tuesday, July 10, 2007 10:02 PM
    Subject: Re: [LSC] LDraw File Format Specification - Comments please


    My comments:

    Charset encoding: technically they're probably DOS (Code Page 437) or DOS/Latin (Code Page 850).  If we're going to pick one, DOS/Latin is the more useful of the two.  I'm not sure if we should mention this or not.  In order to be useful, software would have to do the conversion from that to the current code page.  The likely current behavior (file is always assumed to be in creation computer's code page) is probably more useful, although it introduces problems when files are shared internationally. 

    Another option would be to state that version 1.0 LDraw files are in UTF-8.  That allows for full Unicode support while (mostly) maintaining backwards compatibility.  (Comments with extended characters wouldn't look right in software that didn't tread the LDraw file as UTF-8, but they probably wouldn't break anything.)  The biggest problem with this proposal would be that there would be no way for a new program to be able to tell if a file was actually UTF-8, or DOS character set.  The UTF-8 to UTF-16 conversion routines (provided by unicode.org) would then probably fail if the file contained extended DOS characters, which could be a problem.

    Quads (line type 4) should be required to be in clockwise or counterclockwise order.  In my opinion, the current "recommendation" to do this is insufficient.  The good/bad image in that section should be moved below the text that talks about point order.  Right now it's in a confusing location, since it's unclear what it's referring to.  Additionally, the text flow around it looks bad. 

    I believe that concave quads should be discouraged.  They could perhaps be forbidden, but I definitely think they should be discouraged.

    We should perhaps have some info on overlapping polygons.  At the very least mention that they will render badly in many renderers when they show up in transparent parts.  Additionally, I think it should be forbidden for two co-planar polygons to overlap if they are not the same color, as this produces undefined results in pretty much any renderer. 

    There was some disagreement in the community in my most recent thread on T-junctions here:
    http://news.lugnet.com/cad/dat/parts/?n=6086

    I think my opinion on the matter is fairly clear in that thread, and I'm all for including information in the spec about the problem, along with instructions that they should be avoided when practical.  However, given some of the comments from part authors, I doubt we can outright forbid them in the spec. 

    The "series of pictures" text in the type 5 line section should simply be "picture", since there's only one picture.

    --Travis


    On 7/10/07, William Howard <william at howard-family.fsworld.co.uk> wrote: 
      Hi gang,

      With all the specs that the File Format spec refers to now ratified, it's time to move onto the main event!

      The current file format spec is here - http://www.ldraw.org/Article45.html 
      The current draft spec is here - http://www.ldraw.org/Article218.html 

      I've copied the draft spec into the LSC scratch-pad area - http://www.ldraw.org/Article408.html - and made as many of the 39 changes of my previous email as possible (no idea what item 9 was as it's missing from the email!).  Those I didn't feel able to change without some input from the rest of you I've embedded in red text (and surrounded with [[ and ]] so you can still find them if you print out a copy)

      Don't know if we want to keep all comments under one thread, or start a new thread for each section of the document.

      William

      _______________________________________________
      LSC mailing list
      LSC at ldraw.org
      http://five.pairlist.net/mailman/listinfo/lsc






----------------------------------------------------------------------------


    _______________________________________________
    LSC mailing list
    LSC at ldraw.org
    http://five.pairlist.net/mailman/listinfo/lsc



------------------------------------------------------------------------------


  _______________________________________________
  LSC mailing list
  LSC at ldraw.org
  http://five.pairlist.net/mailman/listinfo/lsc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://five.pairlist.net/pipermail/lsc/attachments/20070730/4b2f3946/attachment.html>