mk_runtool use p_type

[kglotfelty]

With CIAO 4.16 we will have access to a parameter's p_type so together with using the plist method we could replace the line parsing

ciao-contrib/mk_runtool.py

Lines 145 to 152 in 22bbae8

	self.name = toks[0]
	self.type = toks[1]
	self.python_type = self.python_type_map[self.type]
	self.mode = toks[2]
	self.value = toks[3]
	self.minval = toks[4]
	self.maxval = toks[5]
	self.info = toks[6].strip()

with individaul pget's to each of the p_foo properties (p_min, p_max, p_prompt, p_value, and now p_type).

[DougBurke]

I was wondering if this change would simplify the runtool generation, but then immediately forgot about it...

[kglotfelty]

Me too, that's why I created the issue so we don't forget.

I was also wondering if we can do things 'on the fly' instead of pre-generating each individual too/.par file -- but we probably have too many special cases (scripts(*), parnames that start with numbers, etc) that it's probably not worth the effort.)

(*)Though I know Joe's team is working on converting the script to python so maybe that's one less barrier.

[DougBurke]

Hmmm, so this is with CIAOX. When asking for p_min or p_max it is prompting me for an answer, which seems wrong (I tried paramopen with mode set to "r" and "h" and this didn't seem to help).

>>> import paramio as pio
>>> fh = pio.paramopen("dmstat")
>>> print(pio.plist(fh))
['infile', 'centroid', 'median', 'sigma', 'clip', 'nsigma', 'maxiter', 'verbose', 'out_columns', 'out_min', 'out_min_loc', 'out_max', 'out_max_loc', 'out_mean', 'out_median', 'out_sigma', 'out_sum', 'out_good', 'out_null', 'out_cnvrgd', 'out_cntrd_log', 'out_cntrd_phys', 'out_sigma_cntrd', 'mode']
>>> pio.pget(fh, "median")
'no'
>>> pio.pget(fh, "median.p_prompt")
'Calculate median value?'
>>> pio.pget(fh, "median.p_type")
'b'
>>> pio.pget(fh, "median.p_value")
'no'
>>> pio.pget(fh, "median.p_mode")
'h'
>>> pio.pget(fh, "median.p_min")
Calculate median value? (no): 
'no'
>>> pio.pget(fh, "median.p_max")
Calculate median value? (no:) (no): 
'no'

Wondering if it was because this was a boolean, I try with an integer value with min and max set and get even odder behaviour:

>>> import paramio as pio
>>> fh = pio.paramopen("dmstat")
>>> pio.pget(fh, "maxiter.p_min")
'1'
>>> pio.pget(fh, "maxiter.p_max")
Maximum number of iterations (1:) (20): 
'20'

Aha. It seems to query when the value is not set (so for maxiter we have a set minimum value but no maximum, so the query is asking for the value to use). Which is understandable but not useful here.

[DougBurke]

For reference, here's what happens when you have a list of options (here for merge_obs):

>>> pio.pget(fh, "psfmerge.p_min")
'exptime|expmap|min|max|mean|median|mid'

and the environment paths get expanded for you - e.g.

>>> pio.pget(fh, "tmpdir")
'/tmp'

rather than returning "${ASCDS_WORK_PATH}" (not sure if this would be an issue or not)

[kglotfelty]

Humm, not sure if the emails went through

if you open w/ 'rH' (upper case) then you don't get prompted ... you still get the exception if the min/max isn't set

In [3]: ff = pio.paramopen("dmstat", "rH")

In [6]: pio.pget(ff, "maxiter.p_min")
Out[6]: '1'

In [7]: pio.pget(ff, "maxiter.p_max")
/home/kjg/cxcds_param4/dmstat.par: parameter error? : maxiter.p_max
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[7], line 1
----> 1 pio.pget(ff, "maxiter.p_max")

ValueError: pget() Parameter not found

and

You can use 'p_value' to get the parameter value before it has been resolved, so

In [1]: import paramio as pio

In [2]: zz = pio.paramopen("merge_obs", "rH")

In [3]: pio.pget(zz, "tmpdir.p_value")
Out[3]: '${ASCDS_WORK_PATH}'

In [4]: pio.pget(zz, "tmpdir")
Out[4]: '/tmp'

[DougBurke]

Ta. Now get back to not-being-at-work!

[DougBurke]

So, we tried this and it didn't work for 4.16. Let's leave this open in cae we change the paramio behaviour for the next release.