specextract: should it set tabstops when calling dmgroup?

[DougBurke]

In 4.16 I'm hoping that in Sherpa commands like group_counts will automatically ignore the filtered-out data.

Should we allow users to give an energy / wavelength filter and then use that to set up the tabStops. It's a bit annoying to do (need to parse the RMF since we can't guarantee the channel mapping) but it should not be too hard to do. The main difficulty comes in designing a useful and usable API. But hey, what's 2 or more parameters, am I right?

[nplee]

think it's a pairing of tabspec/tabcolumn and/or stopspec/stopcolumn in dmgroup, haven't looked too deeply into it... and technically, we already have an energy parameter available to define the grid that the responses are generated on...

[kglotfelty]

The energy grid the responses are created on isn't necessary the range you want to fit. you want a wider energy range in the responses than you fit so that when you convolve with the RMF the channel edges of the noticed energy range get the contribution from the out-of-noticed-range energies.

I'm also thinking that you don't always just fit one range. I imagine that folks need to fit certain model components to different energy ranges, freeze those values, and then change the energy range to fit a different feature. [I'm thinking cases where say you want to fit the Fe55 line first to freeze it's location and then fit the rest of the spectrum; or fit the low energy first to freeze the absorption.]

Of course if sherpa does it (tabsop mapping/matching) then no worries ... but then why group at all in specextract? Answer: XSPEC. So then is there any benefit for xspec users? Or will this introduce another subtle diff?