Opened 10 years ago

Closed 10 years ago

#597 closed defect (wontfix)

Using custom axis names in coverage constructor WCPS queries

Reported by: olcl Owned by: Piero Campalani
Priority: major Milestone:
Component: petascope Version: 8.5
Keywords: Cc: Peter Baumann
Complexity: Medium

Description

When using the coverage constructor syntax Petascope fails when the axis are not named according to x,y,t convention. The examples below show this.

This works

for c in (CCI_chla_daily) return
encode(
  (float)
coverage histogram over
$px x( 0 : 0 ),
$py y( 0 : 0 ),
$pt t( 0 : 364 )
values  (
add( (c[x(-50:-40), y(45:55),t($pt)]>  -1 ) * c[x(-50:-40),
y(45:55),t($pt)])
/
count(c[x(-50:-40), y(45:55),t($pt)]>  -1 )
)
, "csv")

This fails

for c in (ESA_CCI_v1_0_chlor_a_daily) return
encode(
  (float)
coverage histogram over
$px x( 0 : 0 ),
$py y( 0 : 0 ),
$pday day( 0 : 364 )
values  (
add( (c[x(-50:-40), y(45:55),day($pday)]>  -1 ) * c[x(-50:-40),
y(45:55),day($pday)])
/
count(c[x(-50:-40), y(45:55),day($pday)]>  -1 )
)
, "csv") 

error logs can be found @ http://pastebin.com/xUB7MrND

Change History (9)

comment:1 by Dimitar Misev, 10 years ago

Isn't this fixed in 9.0? I doubt we could go back and fix these things in 8.5 now.

comment:2 by Piero Campalani, 10 years ago

Yes this is fixed in 9.0, but we are still able to fix bugs in release_8.5 , it's still under maintenance afaik. ?

comment:3 by Dimitar Misev, 10 years ago

I believe it's supposed to be a major effort, otherwise we would've fixed this long ago?

comment:4 by Piero Campalani, 10 years ago

I believe this is probably due to a coupe of fetching domain element metadata by type, instead of by name, somewhere in petascope.wcps.server.core.*.

comment:5 by olcl, 10 years ago

how far away are stable release RPM's for version 9? if not far then maybe less important to fix in 8.x but if they are quite a time away then maybe fixing 8.x would be best option (from user perspective)

comment:6 by Dimitar Misev, 10 years ago

Cc: Peter Baumann added

The schedule is for end of this month. So in my opinion it would be better to focus on making 9.0 stable then re-fixing bugs in 8.5

comment:7 by olcl, 10 years ago

i agree, if we are only 2 weeks aways from stable v.9 RPM's then waiting is probably the best option

comment:8 by abeccati, 10 years ago

I think this will be obsoleted by 9.0 (different way of getting axis names afaik) so it should be marked as such and left open for 8.5 maintenance with low priority.

comment:9 by Dimitar Misev, 10 years ago

Resolution: wontfix
Status: newclosed

Cleaning up.

Note: See TracTickets for help on using tickets.