 |
 |
Malcolm may have an opinion at some point. So HISTORY headers end up in the FITS extension presumably unless it's a round-tripped file where the HISTORY was hidden away when the NDF was converted to FITS. I was assuming that all the xxx2ndf routines should propagate history. Tim On Tue, Sep 8, 2009 at 10:13 PM, David Berry <[log in to unmask]>wrote: > 2009/9/9 Tim Jenness <[log in to unmask]>: > > > > > > On Tue, Sep 8, 2009 at 9:49 PM, David Berry <[log in to unmask]> > wrote: > >> > >> 2009/9/9 Tim Jenness <[log in to unmask]>: > >> > > >> > > >> > On Tue, Sep 8, 2009 at 9:06 PM, David Berry <[log in to unmask]> > >> > wrote: > >> >> > >> >> 2009/9/8 tim.jenness <[log in to unmask]>: > >> >> > Provenance tracking can be enabled automatically using an > environment > >> >> > variable (AUTOPROV). No reason why HISTORY propagation can't have a > >> >> > AUTOHIST > >> >> > environment variable. > >> >> > > >> >> > For CONVERT I would be in favour of simply enabling HISTORY when > the > >> >> > file is > >> >> > imported to NDF. > >> >> > >> >> OK. I'll add this issue to my do list. Not sure when it will make its > >> >> way to the top though. > >> >> > >> > > >> > Presumably tweaking CONVERT is easy though? > >> > >> If the plan is to provide a global solution, is it a good idea to jump > >> the gun with hard-wired fixes to particular applications? > >> > > > > I was thinking that CONVERT should make a history component anyhow. > > Principal of least surprise. All the other starlink applications > propogate > > history if history is present but CONVERT is special because it doesn't > > start with an NDF. So if there is any HISTORY at all in the input FITS > file > > does the output NDF have a HISTORY component? > > The fits2ndf prologue says: > > * Implementation Deficiencies: > * - There is no propagation of arbitrary FITS HISTORY headers to > * the NDF's history records. > * - There is no support for FITS World Co-ordinate Systems. > > > First bullet answers your question. Second bullet seems a bit wrong... > > I'll look into enabling default History creation in fits2ndf. > > David > > > > Tim > > >