Mail Archives: geda-user/2012/11/18/20:15:41
On Nov 18, 2012, at 5:37 PM, Kai-Martin Knaak wrote:
> John Doty wrote:
>
>>> It is broken as in
>>> "I-consider-to-pull-because-cvs-is-a-such-a-PITA".
>>
>> I understand the problems of cvs: that's why I use git for my
>> development projects. But the gedasymbols cvs tree isn't a
>> development tree,
>
> Building a viable library _is_ development. The very same reasons that
> make git so valuable for programming projects also apply for library
> building.
Yes, but you don't *develop* in /bin or /lib. You develop in a development tree, and install in /bin or /lib. I think of the gedasymbols tree as an installation target, not a development tree. The development tree has all sorts of extra cruft in it to support development.
>
> For starters, CVS is slow. A simple check for updates takes minutes.
> As the lib grows, restructuring tends to involve more and more files.
> Now, cvs cannot move files. There is no rename of folders, either. You
> are forced to remove and recreate. History gets lost during the
> process.
If you treat the gedasymbols tree as an installation target, not a development tree, these problems disappear. And given that it appears that you'll never convince DJ to change to git, this would appear to be the best option.
>
>
>> I'm getting hungry. We should talk some time about adding
>> spice-prototype attributes to your symbol
>
> IMHO, this only makes sense, if they are compllemented by a lib of
> spice models.
As you know, the primary issue here is legal, and we cannot solve that. Still, defacto standard pinouts are common, and the spice-prototype attributes can reflect that. I'm also considering the possibility of automating the fetch of a model from a URL.
John Doty Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd AT noqsi DOT com
- Raw text -