Re: gEDA-user: gattrib corrupts data

[sdb@xxxxxxxxxx (Stuart Brorson)]

Hmmmm. . . .  This is not good.  Please answer a few questions.

> I am loading schematic repeatedly in gschem and gattrib. It results in
> data corruption. For example, I had attributes
> device=1N4184, value="" and changed them to device=DIODE, value="1N418".
> The same with 2N3904 and BC547 transistors. Then saved.

1.  "Device" is an attribute which is supposed to live on a symbol,
and not be promoted to schematic level.  If you are attaching "device"
at the schematic level, I am not sure if it causes a problem or not.
The routines in libgeda specifically don't promote "device" from
symbols.  I don't know what the behavior is if they find "device" on a
schematic.  

Questions:  Did the value attribute come up correctly? Or were both
device and value gone?  

> Then loaded it into gschem. The attribs weren't there.
> Then loaded into gattrib. They weren't there too.

2.  Did you hit return when you entered the values?  I have seen some
intermittant cases where attrib values don't get entered if you just
type in the value, then do "file->save" without hitting return or
moving to another cell. 

To be on the safe side, I ususally save twice when using gattrib; I
also move the cursor around a little between the two saves.

Yes, this is not a good behavior & I am looking for the problem so I
can fix it.

> I am going to stop using gattrib and use gschem only (after I fix
> the schematic).

With a little care you can continue to use both.  I do, and haven't
been badly bitten. . . .  knock wood . . . .

> It happened to me more times. But I also thought it was I forgot saving.

This is an interesting point.  I am planning on implementing a
lockfile system so that people using gschem and gattrib together won't
get out of synch with their saving.  I will post some ideas about that
in geda-dev at some point in the near future.

Stuart