t.remove: add dual-step removal as in g.remove

[neteler]

At time t.remove removes data without warning and force flag.

There are two disadvantages:

• the user being familiar with g.remove's behaviour will be surprised
• there is no chance to catch the lack of using -rf (of course a user error), leading to the situation:

GRASS 7.1.svn (eu_laea):~ > t.remove lst_aggreg_modis_week@modis_lst_reconstructed_europe_daily
Note: registered maps themselves have not been removed, only the strds

... user's thought in this moment: "Heck, now the strds is gone but I forgot to specify -rf to also get rid of the related raster maps and no easy way to do that now... nor how to figure out their map names!"

I refer to situations where thousands of maps are stored in a mapset with similar names which is common in time series processing.

Solution:

At time it is a bit confusing since -f would be the best flag name (as it is used in g.remove). I don't think that it would be a problem to even redefine the meaning of -f in G7 since t.remove is still rather new. Hence:

• change meaning of -f to "force removal" (without -f only show what would be done, yet don't remove). Advertise also -rf in the message.
• -d Remove all registered maps from the temporal and also from the spatial database (required for actual deletion of files)

[neteler]

Ticket retargeted after milestone closed

[neteler]

Ticket retargeted after milestone closed

[neteler]

Ticket retargeted after 7.0.3 milestone closed

[lucadelu]

Hi, please test this diff to see if it is following the requested behaviour.

Other changes are requested in the testsuite of several t.* modules

[neteler]

The question is how to deal with the flag character to be used...

[lucadelu]

Replying to neteler:

The question is how to deal with the flag character to be used...

Maybe I could apply this changes in trunk and not backport after we could release these in the next stable major release

[martinl]

Ping.

[martinl]

Milestone renamed

[neteler]

Why not have it in 7.2?

[neteler]

Ticket retargeted after milestone closed

[martinl]

What is status of this issue?

[veroandreo]

Replying to martinl:

What is status of this issue?

The behavior described in the ticket is still the current behavior:

• no flag set: removes the strds directly (no warning) -->> I think it would make sense to have a WARNING as in G7:g.remove and a flag to force the removal of strds only

• if only -r flag is set, the user gets: ERROR: The recursive flag works only in conjunction with the force flag: use -rf -->> IMHO, it would be consistent with G7:g.remove to get the list of maps that will be removed plus a WARNING with advice to use both flags (-rf) to actually remove strds and maps, otherwise what is the point of having a flag (-r) that cannot be used alone?

[lucadelu]

Replying to veroandreo:

Replying to martinl:

What is status of this issue?

The behavior described in the ticket is still the current behavior:

• no flag set: removes the strds directly (no warning) -->> I think it would make sense to have a WARNING as in G7:g.remove and a flag to force the removal of strds only

• if only -r flag is set, the user gets: ERROR: The recursive flag works only in conjunction with the force flag: use -rf -->> IMHO, it would be consistent with G7:g.remove to get the list of maps that will be removed plus a WARNING with advice to use both flags (-rf) to actually remove strds and maps, otherwise what is the point of having a flag (-r) that cannot be used alone?

I think this is a behavior change and it could affect some add-ons since -f should be used like in g.remove so to really remove. I think this should released in next major release. I would like to have this behavior:

• without flag it show the warning
• -r to show the maps connected with the temporal dataset
• -f remove temporal dataset (with -r remove maps from temporal framework)
• -d to delete maps from mapset

What do you think?

[veroandreo]

Replying to lucadelu:

I think this is a behavior change and it could affect some add-ons since -f should be used like in g.remove so to really remove. I think this should released in next major release. I would like to have this behavior:

• without flag it show the warning
• -r to show the maps connected with the temporal dataset
• -f remove temporal dataset (with -r remove maps from temporal framework)
• -d to delete maps from mapset

What do you think?

Sorry for my ignorance here, but is a new flag needed?

I was thinking of:

• no flag: shows the warning that to remove strds -f should be used
• only -f: removes the strds, and warns that -rf should be used to remove the maps
• only -r: shows maps in the strds that will be removed and warns that -r and -f should be used to remove the maps
• -rf to remove strds and maps (as now)

[veroandreo]

Since this would imply a change in the behaviour of the flags that is only possible in major releases, we close this ticket as won't fix. See #3362 for such a request for GRASS 8 :)