#49 !g. additions

[Tom Bania]

Colleagues

I continue to methodically migrate my package into GBTIDL.
I have defined the global structure !he3. below in
order to do so.
Of course, I could just invoke this ideosyncratically
as now I do,
but I feel strongly that these globals are of great
utility to anyone
who wants to deal with large data sets.

So, I would ask that the team consider adding these
items to !g.
The first bunch are all the current things that can be
SELECTed
upon. The last have to do with dataio management
within user
written procedures.

Note, too, that I have written some CLASS look alike
functions:
'fi', 'fo', 'gf', 'gn','find' that use/set these
globals... More on this
anon.

;
; Procedure to create IDL He3 structure for SELECT
;
pro he3_struct
;
on_error,2
;
he3 = { he3_struct, $ ;
source:string(replicate(32B,16)), $
projid:string(replicate(32B,32)), $
backend:string(replicate(32B,32)), $
observer:string(replicate(32B,32)), $
bandwidth:0.0D, $
telescope:string(replicate(32B,16)), $ ;
telescop in sdfits
polarization:string(replicate(32B,2)), $ ;
crval4 in sdfits
polarization_num:0L, $
scan_number:0L, $ ; scan in sdfits
procedure:string(replicate(32B,16)), $ ;
obsmodes[0] in sdfits
switch_state:string(replicate(32B,16)), $ ;
obsmodes[1] in sdfits
switch_sig:string(replicate(32B,16)), $ ;
obsmodes[2] in sdfits
obsid:string(replicate(32B,32)), $
frontend:string(replicate(32B,16)), $
sampler_name:string(replicate(32B,16)), $ ;
sampler in sdfits
feed:0L, $ ; feed in sdfits
srfeed:0L, $ ; srfeed in sdfits
feed_num:0L, $
if_number:0L, $ ; from index file
;
;
;
lastrec:0L , $ ; last record# read from
'filein'
recmin:0L, $ ; first rec# for search range
recmax:0L, $ ; last rec# for search range
lastscan:0L, $ ; last scan# gotten from
'filein'
scanmin:0L, $ ; first scan# for search range
scanmax:0L, $ ; last scan# for search range
input_file_name:string(replicate(32B,80)), $
nrecs_in_filein:0L, $ ; total number of records
in filein
output_file_name:string(replicate(32B,80)), $
nrecs_in_fileout:0L $ ; total number of records
in fileout
}
;
defsysv,'!he3',{he3_struct}
;
print,'!he3 structure created'
;
return
end

[Bob Garwood]

Logged In: YES
user_id=978628

Hi Tom,
On the selection issue, we're currently planning on
implementing a FIND (modelled on CLASS FIND) which would be
built on top of SELECT using fields in !g. We already have
pointers in !g to hold the things that FIND would use. We
would need to also provide setsource, setscan, etc, to set
the search parameters for each selectable thing. FIND would
then use any previous set things plus any additional
refinements the user might provide. As things stand now,
this is unlikely to be ready by v1.1, but it should be ready
sometime shortly after that.

On the second set:

!g.line_filein_name
!g.line_fileout_name

Already exist and

nrecords([/keep])

is an existing function, so I don't see the need for your
input_file_name, output_file_name, nrecs_in_filein,
nrecs_in_fileout

I do, though, see the utility of lastrec and lastscan.

Did I miss anything in responding to this?

-Bob

[Tom Bania]

Logged In: YES
user_id=1286618

we are working in parallel. sigh.

when is v1.1 due out?

i take all your points above. missed them. hard to code
when one is only
doing it sporadically. and, i am still catching up on
exactly what you folks
have done....

my only thought as to why the file sizes should be in !g. is
that you then
don't have to constantly invoke nrecords(). sooner or later
this sort of
recomputing everything slows throughput. i have written a
wrapper around
'filein', the CLASS-like 'fi' command which stuffs this file
size into the
!he3.file_size elements.....

insofar as some of the CLASS like stuff is concerned,
perhaps you might
consider adopting/modifying what i have done. perhaps not.
here is an
example:

;+
;
; setpol.pro Sets polarization for a SELECT search via
FIND procedure
; ---------- if pol value not passed, prompts for string
; Syntax:: setpol,"LL" <-- must pass a
string "*" is wildcard
; RR
;
; T.M. Bania June 2005
;-
pro setpol,pol,ask=ask,noinfo=noinfo
;
on_error,2
compile_opt idl2
;
if (n_elements(ask) eq 1) then goto, query
;
if (n_params() eq 0) then begin
print,'Set the Polarization for data
filtering searches:'
print,'Syntax: setpol,"LL" <-- must pass a
string "*" is wildcard'
print,' setpol,/ask <-- prompts for
input
print,' setpol,/noinfo <-- suppress output
return
endif
;
goto, out
;
query: begin
pol=''
print,'Polarization ID= no quotes needed; <CR> means
wildcard'
read,pol,prompt='Input Polarization: '
end
;
out:
;
spol=strtrim(pol,2)
if ( (spol eq "") or (spol eq " ") ) then spol='*'
;
!he3.polarization=spol
;
if (n_elements(noinfo) ne 1) then print,'Polarization for
searches is: ' + !he3.polarization
;
return
end

if you are interested, i have already written in this style:

set <- prompts for stuff invoking .pro's below
setsrc
setid
setpol
setrange <- sets a scan range
setrec <- sets a record range