Simbad query and query_id fail

[Tonione]

I have encountered trouble with simple Simbad queries. I'm on python3.8 on mac OS, with astropy (5.0) and astroquery (0.4.4).

Simple queries such as
from astroquery.simbad import Simbad obj=Simbad.query_object('Kepler-235')
or
from astroquery.simbad import Simbad name=Simbad.query_objectids('Kepler-235')
result both in the same error:
/opt/anaconda3/lib/python3.8/site-packages/astroquery/simbad/core.py:135: UserWarning: Warning: The script line number 3 raised an error (recorded in the errors attribute of the result table): java.lang.NullPointerException

Did anyone encounter the same issue? I

[keflavich]

I can't reproduce this error. It may have been a temporary service issue. Please try again. If it persists, try querying with the cache disabled:

from astroquery.simbad import Simbad
obj=Simbad.query_object('Kepler-235', cache=False)

If that works (and querying with cache enabled does not), you will need to clear your cache.

[Tonione]

Hi, thank you so much for taking the time to look into it. I have no optional argument 'cache' in Simbad.query_object so I cannot check it out. How would proceed to clear the cache otherwise? Thanks again.

[keflavich]

Remove files in your ~/.astropy/cache/astroquery/Simbad/ directory.

This conversation has revealed a bug, though: simbad.query_object() should pass the cache argument on, but it does not.

[Tonione]

Removing cache files worked like a charm, thank you so much!

[aoberto]

Hello, do you think that we could have a maximum duration for the cache ? For instance, like 1 day ?

[aframosp]

I have the same issue as described by @Tonione, this time for simbad.query_objects(). Removing the cache directory helps only momentarily, because the error is happening every ~10 min if I try some queries continiously.

[keflavich]

@aoberto cache expiration has been a wishlist item for a while.

However, the problem reported here seems to be a real one involving caching failed queries; we should never cache failed queries.

[aoberto]

@aframosp Ho yes, I found the origin of the NullPointer, and fix it (I hope) at least it should avoid the wrong behavior, sorry for that.
@keflavich Yes, of course, this is better, but I do not know which one is easier to do

[keflavich]

Not caching failures is both a bit easier and more critical. Even though the upstream part of this issue is solved, I'd like to leave it open until I fix the caching issue.

[keflavich]

#2187 was the solution to this issue. I'm not entirely sure, but this issues may be solved already. Without a clearer MWE (which is hard to make, since the issue is fundamentally upstream), it is hard to test.

@aoberto is there a query we could run against SIMBAD that would reliably fail with exactly this error so we can add a test of how astroquery handles the error?

[bsipocz]

cache timeout has been worked on and we have two abandoned WIP PRs for some fixes. Maybe they can be warmed up and wrapped up?

@ceb8 - would you think #1634 need much more work?