Feature #1564
Implement setter method for GSkyRegion::name()
Status: | Closed | Start date: | 11/02/2015 | |
---|---|---|---|---|
Priority: | Normal | Due date: | ||
Assigned To: | Mayer Michael | % Done: | 100% | |
Category: | - | |||
Target version: | - | |||
Duration: |
Description
Currently, it is only possible to retrieve the name of a GSkyRegion
object. A setter for this property is needed to create several GSkyRegion
objects (e.g. in python) and append them to a GSkyRegions
container.
The following code won’t work without the setter method and will throw an exception:
import gammalib regions = gammalib.GSkyRegions() for i in range(5): region = gammalib.GSkyRegionCircle(83.6,22.0,0.2*i) # region.name("Region_"+str(i)) <-- This needs to be implemented regions.append(region)
Recurrence
No recurrence.
History
#1 Updated by Knödlseder Jürgen about 9 years ago
I agree that such a method should be added.
I do not recall why we have decided that sky regions need to have a name. Maybe we should revisit this issue; at least check why at the time we have implemented the sky regions we decided that each region should have a name.
#2 Updated by Mayer Michael about 9 years ago
I went through the respective issues. Especially in #535, there was a long discussion about the interface of the GSkyRegion
class. In comment 23 you asked about the purpose of the name
member. Pierrick replied he replaced this attribute by a m_type
member. Apparently this did not happen.
I guess after all, we kept this attribute to allow the following access from the container by name: GSkyRegions::operator[](std::string name]
. This makes sense to me. because the user does not necessarily require the id
to access the container element. Drawback: every region the requires a name, different from ""
which is the default.
#3 Updated by Knödlseder Jürgen about 9 years ago
I thought about this a bit more.
We could allow the name to be non-unique. We would then add an integer argument to the operator (that defaults to 0) to access the elements by name. In that way the ith element with a given name can be accessed. At the same time, it does not hurt to leave all names blank (i.e. if one does not want to use the names).
#4 Updated by Mayer Michael about 9 years ago
GSkyRegions.hpp
GSkyRegion* operator[](const std::string& name);
into:
GSkyRegion* operator[](const std::string& name, const int& index = 0);
which would imply non-unique region names.
To implement the changes, we further would need to adapt the following methods:
- GSkyRegion* set(const std::string& name, const GSkyRegion& region);
- void remove(const std::string& name);
- GSkyRegion* insert(const std::string& name, const GSkyRegion& region);
The above functions should also take an additional index as argument, with default value 0, right?
#5 Updated by Knödlseder Jürgen about 9 years ago
I had not recognized before that name
is so widely used.
I’m still wondering about the use cases of the access by name (the reason I’m hesitating is because this now will make the interface overly complex).
#6 Updated by Mayer Michael about 9 years ago
About the use case: I agree the interface would become too complex. I don’t really see the need that every region needs a unique name.
We could instead just keep everything as it is and only make the name
attribute optional. The operator[](std::string)
could then return the first occurrence of the given name. E.g. if every region name is empty, regions[""]
would return the first element.
#7 Updated by Knödlseder Jürgen about 9 years ago
Mayer Michael wrote:
About the use case: I agree the interface would become too complex. I don’t really see the need that every region needs a unique name.
We could instead just keep everything as it is and only make thename
attribute optional. Theoperator[](std::string)
could then return the first occurrence of the given name. E.g. if every region name is empty,regions[""]
would return the first element.
Agree. We may even think about dropping access by name (and also set
, insert
and remove
). I think this was carried over from other classes, but I’m still not convinced that the name
of a region has such a high relevance to merit own access functions (one can always loop over all regions to find a region with a given name). I somehow get the feeling that we make a simple thing complicated.
#8 Updated by Mayer Michael about 9 years ago
- Status changed from New to Pull request
- Assigned To set to Mayer Michael
- % Done changed from 0 to 100
GSkyRegions.i
the operator[](std::string)
is not supported anyway. I hence followed your suggestion and removed the following functions:
set(std::string, GSkyRegion)
insert(std::string, GSkyRegion)
remove(std::string)
operator[](std::string)
contains(std::string)
get_index(std::string)
I recompiled and run make check
to not accidentally break anything. It seems the name
for sky regions was not used in gammalib at all, so everything is works fine.
The changes are available on branch 1564-allow-nonunique-SkyRegions
#9 Updated by Knödlseder Jürgen about 9 years ago
- Status changed from Pull request to Closed
Merged into devel
.