Le mardi 16 juillet 2013, Christophe Geuzaine <<a href="mailto:cgeuzaine@ulg.ac.be" target="_blank">cgeuzaine@ulg.ac.be</a>> a écrit :<br>> But there are ELSETS for physical groups in the file you sent: see at the end of the file...<br>
<br>Ah, yes indeed, I see them now, e.g. by grepping for 'ELSET' rather than 'type'. They are indeed the required Physical entities. Silly me! Sorry.<br><br>> Would you want the additional ELSETS for the elementary entities to be removed?<br>
<br>No, there's no need for that.<br><br>> (We have been trying to rationalize the use of "groups" in Gmsh; eventually we want to make them consistent across formats, and between Gmsh and other tools. The main problem with the old approach was that we always had to renumber the nodes/elements in the output files, and we duplicated elements belonging to multiple physical groups. This made it impossible to have a "non-lossy" write/read cycle, as internally Gmsh could not represent what we saved on disk!)<br>
<br>Thank you very much for the rationale. That makes perfect sense. I had
temporarily forgotten about the possibility of elements having multiple
tags as I'd mostly been using '-format mesh', which doesn't support
that (a limitation of the MEDIT format, defined externally to Gmsh). There are good reasons for occasionally wanting multiple tags, and it's good for formats to suport them. The approach newly adopted in Gmsh does seem to me the best and most flexible way to accommodate that.<br>
<br>Thanks very much.<br><br>Geordie McBain<br><br>