Opened 9 years ago

Closed 9 years ago

#169 closed enhancement (fixed)

us_globalequil - doesn't load experiments

Reported by: demeler Owned by: gegorbet
Priority: normal Milestone: future
Component: ultrascan3 Version: 1.0
Keywords: review Cc:

Description

I tried to load an equilibrium experiment in us_globalequil, but when I clicked on "Load Data" it would only show me 1 velocity experiment when looking in the database, and none on disk, even though I had recommitted a new experiment (demo1_equil_abs) under my name in the CAUMA3 database.

This module should show only equilibrium experiments, not velocity experiments. Now it doesn't show all velocity experiments, and it doesn't show any equilibrium experiments

Change History (11)

comment:1 follow-up: Changed 9 years ago by gegorbet

  • Resolution set to invalid
  • Status changed from new to closed

According to us_manage_data, the demo1_equil_abs experiment has AUC data, but no EDITS in the database, for the demeler investigator. US_GlobalEquil, like other analysis programs, will only present edits. I don't see any failure of us_globalequil to load data (edits) - I've loaded several over the last few days.

US_ManageData shows only 4 EDITS for demeler in CAUMA3, all are from the demo1_intensity runID.

If there is a defect (perhaps in saving edits to the DB), it is not in us_globalequil's ability to load data.

comment:2 in reply to: ↑ 1 Changed 9 years ago by demeler

  • Resolution invalid deleted
  • Status changed from closed to reopened

Replying to gegorbet:

According to us_manage_data, the demo1_equil_abs experiment has AUC data, but no EDITS in the database, for the demeler investigator. US_GlobalEquil, like other analysis programs, will only present edits. I don't see any failure of us_globalequil to load data (edits) - I've loaded several over the last few days.

US_ManageData shows only 4 EDITS for demeler in CAUMA3, all are from the demo1_intensity runID.

If there is a defect (perhaps in saving edits to the DB), it is not in us_globalequil's ability to load data.

In addition to the fact that the ticket 163 has not been resolved yet - which may be the reason for the lack of edits in the database - this program should never load velocity data. Such data should not be shown. Only equilibrium data should be shown in the data loader. Can you filter on equil experiments? Similarly, velocity programs should not display data marked as equilibrium data during the us_convert process.

comment:3 Changed 9 years ago by demeler

  • Status changed from reopened to new

comment:4 Changed 9 years ago by bdubbs

  • Keywords review added

I think this is related to the us_edit problem. I'm going to mark it 'review' for now, because after fixing #163, it appeared to work for me.

comment:5 Changed 9 years ago by demeler

Update: The database loading of equil edits is working now after Bruce fixed the editing routine. I tried it with a brand new dataset and it worked flawlessly. The issue that remains is that the program should only show equilibrium data to the user, not velocity. Once you have lots of experiments in the database, you want to pre-filter on the only applicable type, which is equilibrium.

comment:6 follow-up: Changed 9 years ago by gegorbet

The originally reported defect was, in my opinion, invalid. There was no problem loading equilibrium data, if it existed.

The newly added enhancement item of filtering on data type requires some thought. To implement this, at this time, would be quite disruptive. The data type is not carried in the DB edit records, nor in local edit XML files. So, either the edit records would need to be enhanced both in the DB and locally; OR the data loader would need to actually read AUC data (where data type currently resides) in preparing the list. This later would likely significantly slow down the *preparation* of a list, changing it into a *load* of *all* data sets. The former (adding type to DB edit records and local edit XML) would require changes to DB tables, us_convert, and us_edit; plus, of course, to the data loader itself.

Note that there is already a text filter on description available. If you type "eq" in the data loader Search field, you will likely reduce your list to equilibrium data sets.

comment:7 in reply to: ↑ 6 ; follow-up: Changed 9 years ago by demeler

Replying to gegorbet:

The newly added enhancement item of filtering on data type requires some thought. To implement this, at this time, would be quite disruptive. The data type is not carried in the DB edit records, nor in local edit XML files.

That's not correct - it should be readily available. The type is carried both in the DB and in the XML - if it is not then that's an error Dan needs to address. This is a qualifier for the <experiment type=equilibrium> xml tag and should be available when loading from either DB or disk. So all you would need to do is to parse the experiment type from the XML or the appropriate DB record - Dan can tell you which one that is. Similarly, this filtering should be done in velocity experiment analysis methods as well - of course there you would filter on type=velocity. type=other should be ignored by both for now. In that category fall things like rotor calibration runs.

comment:8 in reply to: ↑ 7 Changed 9 years ago by gegorbet

Replying to demeler:

Replying to gegorbet:

The newly added enhancement item of filtering on data type requires some thought. To implement this, at this time, would be quite disruptive. The data type is not carried in the DB edit records, nor in local edit XML files.

That's not correct - it should be readily available. The type is carried both in the DB and in the XML - if it is not then that's an error Dan needs to address. This is a qualifier for the <experiment type=equilibrium> xml tag and should be available when loading from either DB or disk. So all you would need to do is to parse the experiment type from the XML or the appropriate DB record - Dan can tell you which one that is. Similarly, this filtering should be done in velocity experiment analysis methods as well - of course there you would filter on type=velocity. type=other should be ignored by both for now. In that category fall things like rotor calibration runs.

The type is carried in the XML. It is not carried in the edit DB records. It could be added in those tables; or logic could be changed to pick up the experiment ID and do a separate read of the experiment record to get the type value. A type filter would need to be added to the data loader and probably all analysis methods changed to pass the appropriate filter value.

Is this something we want to do as part of Beta? Would it be better to make it a "Future" item?

comment:9 Changed 9 years ago by bdubbs

The experiment type is in the DB experiment table:

type ENUM('velocity', 'equilibrium', 'other') NULL,

perhaps it should be changed to NOT NULL DEFAULT 'other'

The type is specified in the stored procedure call new_experiment and updated in update_experiment. It is returned in get_experiment_desc, get_experiment_info and get_experiment_info_by_runID.

comment:10 Changed 9 years ago by gegorbet

  • Milestone changed from beta to future
  • Status changed from new to assigned
  • Type changed from defect to enhancement
  • Version set to 1.0

Filtering by experiment type (in us_globalequil's case, "equilibrium") has been added to the data loader and analysis programs. For the DB, an additional get_experiment_info call is needed, slowing the appearance of the edit list. This slow-down may be partially resolved in the future with an enhancement to the all_editedDataIDs procedure API (#174).

comment:11 Changed 9 years ago by dzollars

  • Resolution set to fixed
  • Status changed from assigned to closed
Note: See TracTickets for help on using tickets.