[QGIS-Developer] timezonefinder discussions

[C Hamilton]

First I want to thank everyone for their input in this discussion, but it
did get off track. Originally, I asked whether my Date/Time Tools plugin
<https://github.com/NationalSecurityAgency/qgis-datetimetools-plugin> I
created for our users could be added to plugins.qgis.org. It includes the
timezonefinder library which is ~42 Mb in size and exceeds the maximum size
of a plugin for inclusion in the QGIS plugin repo.

Note that I have a very good reason for making sure that every one of my
plugins are self contained primarily because my users don't manage their
machines and consequently cannot install extra python libraries which is
what timezonefinder is so it has to be included with the plugin.

It was suggested to not use timezonefinder, but just use the time zone
polygon layer. That way I wouldn't need to bundle an extra python library.
I wanted to test if this would be as fast as timezonefinder. Here is where
the conversation got side tracked because I created the 10,000 random point
layer for the purpose of testing how quickly "single point" lookups could
be done using timezonefinder vs. some QGIS point in polygon look up. It was
NOT about finding the time zones for the 10,000 points all at once. The
10,000 points were just a data source for timing single point lookups so
although "Join Attributes by Location" does a great job of identifying all
10,000 points at once that was NOT what I was trying to accomplish and I
don't think that would be the proper solution for looking up single points,
such as displaying the time zone in an info box as the mouse moves over the
QGIS canvas. On each motion event, a single point lookup is done.

1. Currently timezonefinder works great for my users. Although the plugin
is ~50Mb if I were to removed timezonefinder library and just use the time
zone polygon layer it would increase to ~100Mb. This could be an extra
download, but I like to make it easy for my users and just include it.

2. Using the time zone gpkg polygon layer is slightly more accurate, but
not by much.

3. So far I have not found that single point lookups in the time zone
polygon layer to be faster than the timezonefinder lookup. This may be
because I am currently developing on a Windows AWS instance and the point
in polygon lookups may be slow because the time zone gpkg layer resides
somewhere on the file system.

4. There is merit in using the time zone polygon gpkg layer. There are some
other algorithms that could be implemented that would benefit from it.
Right now I only have a Date, Time, and Time Zone conversion tool and
either method would work fine for this.

5. I still have a lot to think about including implement some astronomical
algorithms of the sun and moon. The plugin already implements some sun
related information.

I really appreciate everyone's comments. They have given me some new ideas.
For my users, timezonefinder does what we need, but it is probably not the
best method for the QGIS community especially with the plugin size
limitation.

I really don't want to take up any more of your time especially you QGIS
developers who have so many other duties. We can end the discussion here.
My priority will always be for my users, but I will try to make the plugin
useful to the community as well as long as it doesn't take me too much more
time to satisfy the plugin size restriction.

Thanks for all you do,

Calvin

On Wed, Feb 10, 2021 at 3:48 AM Richard Duivenvoorde <rdmailings at duif.net>
wrote:

> Hi Chris,
>
> I see all this discussions on the list (going round and round....).
>
> Wasn't the conclusion that:
> 1) you just use your first version (which included that data, and was a
> little big....)
> 2) put it on plugins.qgis.org (via Tim, as uploading it would probably
> fail because of size)
> 3) we keep the 'max size' of plugins small, to not get overloaded with
> super heavy plugins?
>
> To not taking up too much energy from you (or all mail thread responders)
> :-)
>
> Or do *I* miss some information?
>
> Regards,
>
> Richard Duivenvoorde
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20210210/ee428358/attachment.html>