First of all, thanks guys for interesting replies to my original post!

I really think by attracting more non-software to the OpenMoko project (and also to projects like OScar) could be the trigger for creating non-software developing communities. Many engineers are pretty tired of dealing with the messy CAx/"PxM" situation, and could probably sacrifice a high salary (like software developers started to do in the beginning of the 90's. It would probably take lots of resources (the resemblance with developing the Linux kernel is probably pretty accurate), but if just a small fraction of the amount spent on proprietary CAx/PxM systems could be spent on open source / open standard development, I think it could be possible!


Michael Shiloh wrote:
I just returned from ETech, the emerging technology conference, and some of us discussed this problem. (In fact, the conversation started because of the Openmoko release of Freerunner CAD files.) All programming languages have a common source code form: ascii text. In contrast, there is no such standard for mechanical drawings or schematic drawings.

What about STEP? It would be very good if CAD data could be human readable, and thus being able to use version control systems like Subversion and Git in a similar way to source code.
Similarly, there is no single standard for image data. There are a multitude: JPEG, TIFF, BMP, etc. but this is not as bad a problem because tools exist to convert from almost any format to any other format.

But the big difference between these image data and CAD data is that the specifications are open for the image data, but not for the CAD data. Please correct me if I am wrong.
Since it's unlikely nor desirable that a single standard will emerge for mechanical or schematic capture, we came up with the idea of putting out a call to the open source community to develop tools to convert from from one format to another.

Some of this exists already, for example in Blender's ability to import from a number of different file formats. But we would like to see a project dedicated to creating a collection of tools to do this.

Could "CAx functionality" be built on top of Blender?
Comments?

Mark Arvidson wrote:
    The reason why I joined list is that I found the CAD files of the
    Neo1973 at the openmoko.com <http://openmoko.com> site. It is great
    that these files are
available, even if they are in a proprietary data format (the ProE, not
    the STEP one), and very likely developed using a proprietary CAD
    application.


Developed with Pro/E if they are Pro/E files.

    Personally, I have a software development company, in which I have
    been developing tailored software for my clients, mostly with open
    source technologies, and sometimes with, at least well known, closed
source techs like .NET . Therefore, it was a shockening moment when I
    during 2007 did a consultancy project for an industrial company
producing water taps. It meant going back at least ten years in time, back into proprietary hell! Proprietary systems (including proprietary
    interfaces between systems) and proprietary data. CAD was done with
proprietary software, often with more than one CAD system, resulting in
    incompatible, binary-only data.


Having come from inside that industry, I have to agree. A very big mess.

    So my questions for the OpenMoko community are:

    * Does there exist any "usable" open source CAD systems? (Is perhaps
    Open CASCAE a viable semi-open http://www.opencascade.org/ option?)


Open CASCADE may be an option for a kernel. I do not know much about it, except it has been used for numerical finite analysis more than modeling. It could probably do the work, but perhaps a bit slowly?

CAD is a very complex subject. There are many different solutions for mechanical design, but only 3-4 for parametric controlled modeling engines, and they are all high-dollar proprietary software packages with extremely rigid licenses. To date, I know of no OSS projects to try to create a parametric modeling engine. The original and on-going development of the 3D parametric modeling engines (such as ACIS or Parasolid) has taken many, many millions of dollars, so is a major OSS undertaking, perhaps similar to the Linux kernel. Blender has a sort of add-on parametric plugin, but it is quite limited. Alibre Design Xpress is free, but proprietary.


    * Is it possible to use a human readable format for CAD data? (Is
perhaps STEP enough for development, or just a format for interchange
    between different CAD applications?)


STEP is a good, complete standard format that I believe all major packages support well. If I were going to create an OSS 3D modeler with human readable format, STEP is a good way to go. IGES is/can be a human readable format, but you lose the parametrics with IGES. The format was design to drive CNC machines, so is more about the model exterior than anything else. It's original design was based on punch-cards, so is very heard to read by humans directly.

There are xml formats, too, but none are really very standardized.

    And the most important question:

    * Is the OpenMoko community interested in  using open source tools
    (possibly together with a human readable format) for developing
    "non-software" parts?


I think this product has primarily attracted software engineering types. We should get the word out to non-software people and increase the audience for that question.

--Mark Arvidson


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

_______________________________________________
OpenMoko community mailing list
community@lists.openmoko.org
http://lists.openmoko.org/mailman/listinfo/community


</div>

_______________________________________________
OpenMoko community mailing list
community@lists.openmoko.org
http://lists.openmoko.org/mailman/listinfo/community

Reply via email to