kicad-developers team mailing list archive

[jp charras <jp.charras@xxxxxxxxxx>]

Le 20/09/2017 à 16:33, Maciej Sumiński a écrit :
> On 09/16/2017 09:32 AM, jp charras wrote:
>> Le 16/09/2017 à 07:28, Russell Oliver a écrit :
>>> Hi, 
>>>
>>> Jean-Pierre: can you send me a link to the files you tested it on? 
>>>
>>> Also for opening either an Eagle schematic or board file, the intended workflow starts within the
>>> KiCad launcher. File > Import Project > Eagle Cad. This launches both Eeschema, Pcbnew for the
>>> matching files and then synchronises the timestamps across the files. 
>>>
>>> Regards
>>> Russell
>>>  
>>
>> I found the issue!
>>
>> The crash is due to the fact certainly the schematic Eagle plugin is missing a switch to the "C"
>> locale (adding a "LOCALE_IO dummy;" declaration before reading the file), so the floating numbers
>> create issues in countries that use the comma as floating number separator (like in France).
>>
>> However, when parsing a file, the Eagle plugin should show an error and abort the load process,
>> without crashing Eeschema.
> 
> I rebased the changes on the current master branch and pushed to my
> repository [1]. There are also some modification, that hopefully make
> the branch ready to merge:
> - code formatting
> - added LOCALE_IO object (JP, would you confirm the problem you
> experienced is fixed?)
> - separated open and import functions to resemble the pcbnew approach
> - added exception handler for the import function, so any errors should
> result in an error message rather than segfault
> 
> Regards,
> Orson
> 
> 1. https://code.launchpad.net/~orsonmmz/kicad/+git/kicad/+ref/eagle-import
> 

Thanks Orson,

I confirm the locale issue is fixed.

After tests with 2 schematics I found a rounding issue in coordinates (especially wires and pins):
Many coordinates are very near (1 mil, therefore one Eeschema unit) the 50 mils grid but not exactly
on grid (for instance 499 instead of 500 mils)
The a few values I checked are smaller (one mil) than the grid coordinate, so I am thinking this is
a rounding issue.

-- 
Jean-Pierre CHARRAS