[Osmf-talk] Wrap-up email about an OSMF Tasking Manager Was:Re: Why not an OSMF Tasking Manager ?

Augustin Doury xhemp at mykolab.com
Sun Dec 23 12:00:36 UTC 2018


Disclaimer: english is not my native language, please don't focus on
specific vocabulary choices 

Hi all, 

Thanks for your different answers ! 

 	* QUICK REMIND

Firstly I would like to remind quickly the 4 reasons why I think "OSMF
should propose an independant and robust Tasking Manager alternative to
the OSM community".
The detailed arguments are available in the first email of this thread. 

 	* Tasking Managers are used by thousands of mappers and concern
millions of edit, the software seems to be an inescapable way of
contribution for many mappers
 	* propose a TM instance which is not necessarily linked to
Humanitarian stuff (but could be)
 	* propose a TM instance which is stable and independant of the Area of
Interest
 	* propose a TM instance where OSM contributors can contribute
independently of any official organisation

 	* ABOUT THE ISSUE AROUND TECHNICAL SKILLS TO RUN AN INSTANCE

@Blake : thanks to propose your technical skills about TM install, as
you said "under the auspices of the OSMF and appropriate Working Group".
Happy to know that an OSMF TM instance could make sense for you ! 

 	* ABOUT THE COUNTER-ARGUMENT "THERE ARE PLENTY OF EXAMPLES OUT THERE
FOR RUNNING A TM OUTSIDE THE HOT DOMAIN".

Thanks Steven and Ralph for your answers. Despite all this instances,
I'm still convinced of the initial proposition for an OSMF TM.
=> my initial email mentioned that : "_They are many existing instances
[link [1]], but most of them are "country based", are not always
up-to-date (the migration from v2 to v3 is not easy), and do not
necessarily offer a "guarantee of service"_. 

If needed, here are more details to illustrate that. 
20 TM instances are "declared" on the Wiki: 

 	* 9 instances are running TM version 3 while 8 are running TM version
2
 	* 3 doesn't seem to work
 	* 15 are country-based
 	* 5 focus on humanitarian&development or environment or historical map
 	* 0 propose an alternative which could be compatible with arguments
above (cf "Quick Remind" point) n°2, 3 and 4 in favor of "OSMF should
propose an independant and robust Tasking Manager alternative to the OSM
community". But it's always possible to debate on arguments !

I think it's interesting that there are at least 15 country-based TM
(Canada, Colombie, Finlande, Portugal, Indonesia etc.) which could be
interpreted as a wish by countries communities to have a local
independent instance, and also probably an instance where people can
"get creation rights (like software is not designed for anyone to create
a project) [...] within a short time" as Severin explained.
IMO these observations are in favour of an OSMF independant and
up-to-date TM instance. As a matter of fact, a lot of mappers from local
communities can't have access to a country-based local instance (because
their country doesn't have one) and so don't have any other choice than
choose an instance focused on thematics (hum/dev, environment etc.)
and/or run by a specific organization for which TM users are indeed
considers as official volunteers. 

 	* ABOUT  "TM TECHNICAL LIMITS AND POTENTIAL RESOLUTION" RAISED BY
SEVERIN

Thanks Severin to show your interest and to propose to participate in
this topic. You accent on some issues which could not be ignored if OSMF
decides to run a TM instance, particularly (as I understood): 

 	* multiple instance implies : how to ensure a minimum coordination
between them ? (ex : avoid to create two similar projects on 2 different
instances) >>> this problem already exists and would not be specific to
an OSMF TM instance.
 	* how to maintain a "generic" fork of the TM Github main repo ? that
should not be complicated if we just personalize main text areas.
 	* how to deal with the TM "project manager" status, given that TM "is
not designed for anyone to create a project" >>> IMO, the must
challenging point to discuss further
 	* how to deal with "organized edits" ? running an OSMF TM would be an
"an additional workload" for OSMF. We must think about a "test period".

 	* WHAT'S NEXT ?

With Blake and Severin it's like at least 3 people are interested in
working/help on this subject, maybe the begining of a WG ?
I think it could be so valuable to have: 

 	* more inputs from people on this list
 	* an idea of what could be the conditions to realize this project
"under the auspices of the OSMF and appropriate Working Group"

Have a good day ! 

Augustin Doury 

PS : because my message was rejected by the list (too big), I've deleted
thread's older emails from below  

----- 

COPY/PAST from Severin email which is only readable outside the mailbox
and I'm not sure everybody got it >> Le 2018-12-21 22:57, Severin Menard
a écrit : 

(Original en français, automatic translation by Deepl with slight
corrections below) :
https://mail.protonmail.com/eo/message/jup9rQTLk_Fm56vD1vZoVILs1K-oN7h-B9BbEVMh6DKnJLjjrzTyVgmn5KfQGYHCJblrxMTIFPyuBZQdXIFoKg==

I think Augustin is totally aware of the existence of different
instances of the Tasking Manager (TM). In the past, he has even
administered a server hosting a TM instance.                            
     

The creation of an OSMF instance of the Tasking Manager is an idea to be
explored, but it is important to be aware that it will not de facto
solve some of the concerns encountered in the implementation and
coordination of the various projects present in the different TM
instances.                                                              


The main current limitation of the TM remains the lack of a clear vision
of the projects created. True, version 3 of the TM includes a mapping in
the form of punctuals categorized according to their degree of priority,
but this is not enough to visualize the spatial extension of projects
and only concerns those specific to each instance. There is currently no
software development to recover the spatial extensions of each project
(which the TM has long produced in GeoJSON format) from the various
deployed instances of the TM (at least those known) in the form of a
geographical layer that would replace the silo point mapping of version
3. It would show the spatial extensions and also share some information:
host authority, date of creation, date of last contribution, %
completeness, typology (e.g. buildings only, detailed baseline data,
road network, residential areas, etc.) and a link to the project. This
mash up of existing inter-instance projects would be integrated into the
steps of creating a new project on each of the instances integrating
this new software development, which would allow any creator to ensure
that they do not duplicate an existing project in any part, or at least
to be aware of it. 

This is a development that the Foundation could consider funding in case
the NGO HOT US Inc, which has carried out the latest developments of the
TM, does not wish to do so.  

However, it would not prevent the implementation of a TM managed by the
OSMF, if its members so wish. This body would de facto represent the
most generic a priori body of the TM, but I do not believe that it would
be more official than the existing bodies and even less that it could
become the only available TM, because 1 basically a free service is more
efficient when it is replicated, 2 this would go in the opposite
direction of the decentralising will of the OSMF and 3 the groups or
structures that implement their own TM are committed to it.             
                                                                        
                                    

In my own experience as administrator of users of an instance of the
Tasking Manager, the problem for anyone who wants to be able to create a
project is not to find an instance, but to get creation rights (like
software is not designed for anyone to create a project) and often
within a short time. These OSM contributors naturally go to the first
instance they naturally think of (that of HOT US Inc or on a national,
regional, thematic basis, etc.) and to a second instance if they could
not have creative rights in the first. It would then be up to the OSMF
to find a way to satisfy requests for additional fees in its
proceedings. 
One solution could be that this instance could be open to any OSM
contributor without specific rights for the creation of new projects,
but this would also require a TM software development which currently,
unless I am mistaken, does not allow this. 

This does not mean that the creation of a TM project via an open
instance is not without constraints. Since the TM involves in the vast
majority of its projects managed editions, for which a guideline was
recently defined by the DWG and approved in November by the OSMF board,
logically each project creator of the OSMF instance will have to create
a wiki page explaining the modalities of the managed editions of the
project. This instance of the OSMF would also necessarily constitute an
additional workload for the DWG and therefore a possible brake. These
questions and constraints would probably require starting with a
time-limited test period followed by an evaluation. I would be happy to
participate in this topic.                      

Severin 

 

Links:
------
[1]
https://wiki.openstreetmap.org/wiki/OSM_Tasking_Manager#Operational_installations_of_the_Tasking_Manager
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmf-talk/attachments/20181223/1b512181/attachment.html>


More information about the osmf-talk mailing list