Re: [QGIS-Developer] Backporting and 3.10 -- a reminder

[Peter Petrik]

Understood, thanks for (re-)clarification and doing the hard work there.
P.

On Thu, Sep 17, 2020 at 1:21 AM Nyall Dawson  wrote:

> Hi list,
>
> Some quick reminders about how the current 3.10 LTR should be handled
> during the 3.16 bug fixing round. My apologies in advance about the
> tone here, but I feel strong language IS required in order to
> reinforce the severity of this message!
>
> 1. At this point in time, the 3.10 LTR is mature and we need to take
> EXTREME care when backporting fixes to it. Only low risk, high-impact
> bug fixes should be backported at this stage. Minor bug fixes or high
> risk changes should NOT be backported, and instead should be included
> in 3.16 only. !!!THE RISK OF REPUTATION HARM TO QGIS
> CAUSED BY INTRODUCING REGRESSIONS IN 3.10 CANNOT BE
> UNDERSTATED!
>
> 2. If I wasn't clear enough in (1): DON'T BE SELFISH!. It doesn't
> matter how important a minor/high risk bug fix is to YOU/YOUR customer
> -- think of all the other millions of users of QGIS who will be
> impacted if you introduce a regression!!
>
> 3. Matthias Kuhn and I are the acting gatekeepers to the LTR branch.
> PSC haven't yet made these roles official, but they DID give us
> direction to act in this role for the present. DO NOT MERGE TO LTR
> WITHOUT MATTHIAS OR MY APPROVAL. (If you disagree with this, PLEASE
> raise objections on the PSC list. But until further direction from PSC
> this is the status quo and we'll both be carrying out this role).
>
> 4. In order to protect the stability of LTR, Matthias and I introduced
> the one-month-delay for backport inclusion policy a few cycles ago.
> This has proved very valuable, and has prevented numerous serious
> regressions from ever touching the LTR branch. (win!) We'll manage
> this process by tagging 3.10 backports with a "NOT FOR MERGE" label,
> and then as soon as the backport is approved by someone (which
> **doesn't** have to be Matthias or me, it just has to be someone who
> isn't the original developer) then the backport will be moved to the
> "collected pending backports PR" (eg
> https://github.com/qgis/QGIS/pull/38748). We're doing all the hard
> work here to keep things easy for you and to keep the PR queue
> flowing.
>
> 5. The only exceptions to the one-month-delay policy are:
> - crash fixes or serious data corruption fixes
> - fixes which repair regressions which have been introduced in an LTR
> patch release
> Backports meeting these criteria can be merged immediately without the
> time delay, but again, Matthias and I will manage this.
>
> (and now for the carrot): IMO 3.10 is a great release, and in a very
> good state. I'm proud of how mature and stable 3.10 is, and how stable
> it is. Let's keep it that way :D
>
> Nyall
> ___
> QGIS-Developer mailing list
> QGIS-Developer@lists.osgeo.org
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

[QGIS-Developer] Backporting and 3.10 -- a reminder

[Nyall Dawson]

Hi list,

Some quick reminders about how the current 3.10 LTR should be handled
during the 3.16 bug fixing round. My apologies in advance about the
tone here, but I feel strong language IS required in order to
reinforce the severity of this message!

1. At this point in time, the 3.10 LTR is mature and we need to take
EXTREME care when backporting fixes to it. Only low risk, high-impact
bug fixes should be backported at this stage. Minor bug fixes or high
risk changes should NOT be backported, and instead should be included
in 3.16 only. !!!THE RISK OF REPUTATION HARM TO QGIS
CAUSED BY INTRODUCING REGRESSIONS IN 3.10 CANNOT BE
UNDERSTATED!

2. If I wasn't clear enough in (1): DON'T BE SELFISH!. It doesn't
matter how important a minor/high risk bug fix is to YOU/YOUR customer
-- think of all the other millions of users of QGIS who will be
impacted if you introduce a regression!!

3. Matthias Kuhn and I are the acting gatekeepers to the LTR branch.
PSC haven't yet made these roles official, but they DID give us
direction to act in this role for the present. DO NOT MERGE TO LTR
WITHOUT MATTHIAS OR MY APPROVAL. (If you disagree with this, PLEASE
raise objections on the PSC list. But until further direction from PSC
this is the status quo and we'll both be carrying out this role).

4. In order to protect the stability of LTR, Matthias and I introduced
the one-month-delay for backport inclusion policy a few cycles ago.
This has proved very valuable, and has prevented numerous serious
regressions from ever touching the LTR branch. (win!) We'll manage
this process by tagging 3.10 backports with a "NOT FOR MERGE" label,
and then as soon as the backport is approved by someone (which
**doesn't** have to be Matthias or me, it just has to be someone who
isn't the original developer) then the backport will be moved to the
"collected pending backports PR" (eg
https://github.com/qgis/QGIS/pull/38748). We're doing all the hard
work here to keep things easy for you and to keep the PR queue
flowing.

5. The only exceptions to the one-month-delay policy are:
- crash fixes or serious data corruption fixes
- fixes which repair regressions which have been introduced in an LTR
patch release
Backports meeting these criteria can be merged immediately without the
time delay, but again, Matthias and I will manage this.

(and now for the carrot): IMO 3.10 is a great release, and in a very
good state. I'm proud of how mature and stable 3.10 is, and how stable
it is. Let's keep it that way :D

Nyall
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Adding Field Calculator Functions - Best Practices / Avoiding Name Conflicts

[Nyall Dawson]

On Thu, 17 Sep 2020 at 01:55, C Hamilton  wrote:
>
> Greetings,
>
> Because of a user request I have been starting to expose Lat Lon Tools 
> coordinate format conversion functions. I found out today that if you 
> accidentally call one of your functions the same name as one of the existing 
> functions, it causes QGIS to crash immediately. It got me thinking about the 
> best practices for releasing field calculator functions. As other developers 
> create their own field calculator functions there is the potential for major 
> problems. I would say that if there is a name conflict it shouldn't crash 
> QGIS.
>
> These are the function names I currently have in Lat Lon Tools.
>
> mgrs, mgrs_100km, mgrs_east, mgrs_north, mgrs_to_point
>
> Here are additional function names I have added that have not yet been 
> uploaded to the QGIS plugin repo and more are on the way
>
> utm, utm_epsg, utm_hemisphere, utm_to_point, utm_zone, dms, ddmmss
>
> I like simple function names, but because of concerns that there could be 
> name conflicts with other QGIS plugin developers I am wondering if a plugin 
> name prefix should be added such as ll_mgrs for the Lat Lon Tools mgrs 
> conversion function.
>
> What are your thoughts?

These sound like really useful functions of widespread appeal. Have
you considered implementing them in QGIS itself? I'd be happy to
assist!

Nyall



>
> Thanks,
>
> Calvin
> ___
> QGIS-Developer mailing list
> QGIS-Developer@lists.osgeo.org
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Which qgis-dev in osgeo4w?

[Paolo Cavallini]

Thanks a lot Juergen. So apparently the website is slightly wrong.
Cheers.

Il 16/09/20 17:46, Jürgen E. Fischer ha scritto:
> Hi Paolo,
> 
> On Wed, 16. Sep 2020 at 17:13:52 +0200, Paolo Cavallini wrote:
>> https://www.qgis.org/en/site/forusers/download.html
>> that the nightly is called qgis-dev-full
>> whereas in
>> http://download.osgeo.org/osgeo4w/x86_64/versions.html
>> it is called qgis-full-dev, but it does not seem up to date (2.99.0-2);
>> qgis-dev, 3.15.0-54 might be the correct one. Should the website be updated?
>> BTW, how can an user know which revision has been compiled, before
>> actually installing it?
> 
> qgis-full-dev doesn't contain files.  It's just a list of useful, but not
> required qgis dependencies (eg. python extensions for plugins or gdal-ecw).
> And it doesn't necessarily change which each release.
> 
> The result of the nightly builds is qgis-dev / qgis-rel-dev / qgis-ltr-dev.
> And https://www.qgis.org/en/site/forusers/alldownloads.html has the SHAs of
> those builds.
> 
> The standalone installers are made from the "full" packages, ie. contain more
> packages than the bare minimum.
> 
> 
> Jürgen
> 
> 
> PS: the ongoing master build is for 8929e74a57, so it will have 1e715580af69.
> 

-- 
Paolo Cavallini
www.faunalia.eu - QGIS.org
training, support, development on QGIS, PostGIS and more



signature.asc
Description: OpenPGP digital signature
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

[QGIS-Developer] Adding Field Calculator Functions - Best Practices / Avoiding Name Conflicts

[C Hamilton]

Greetings,

Because of a user request I have been starting to expose Lat Lon Tools
coordinate format conversion functions. I found out today that if you
accidentally call one of your functions the same name as one of the
existing functions, it causes QGIS to crash immediately. It got me thinking
about the best practices for releasing field calculator functions. As other
developers create their own field calculator functions there is the
potential for major problems. I would say that if there is a name conflict
it shouldn't crash QGIS.

These are the function names I currently have in Lat Lon Tools.

mgrs, mgrs_100km, mgrs_east, mgrs_north, mgrs_to_point

Here are additional function names I have added that have not yet been
uploaded to the QGIS plugin repo and more are on the way

utm, utm_epsg, utm_hemisphere, utm_to_point, utm_zone, dms, ddmmss

I like simple function names, but because of concerns that there could be
name conflicts with other QGIS plugin developers I am wondering if a plugin
name prefix should be added such as ll_mgrs for the Lat Lon Tools mgrs
conversion function.

What are your thoughts?

Thanks,

Calvin
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Which qgis-dev in osgeo4w?

[Jürgen E . Fischer]

Hi Paolo,

On Wed, 16. Sep 2020 at 17:13:52 +0200, Paolo Cavallini wrote:
> https://www.qgis.org/en/site/forusers/download.html
> that the nightly is called qgis-dev-full
> whereas in
> http://download.osgeo.org/osgeo4w/x86_64/versions.html
> it is called qgis-full-dev, but it does not seem up to date (2.99.0-2);
> qgis-dev, 3.15.0-54 might be the correct one. Should the website be updated?
> BTW, how can an user know which revision has been compiled, before
> actually installing it?

qgis-full-dev doesn't contain files.  It's just a list of useful, but not
required qgis dependencies (eg. python extensions for plugins or gdal-ecw).
And it doesn't necessarily change which each release.

The result of the nightly builds is qgis-dev / qgis-rel-dev / qgis-ltr-dev.
And https://www.qgis.org/en/site/forusers/alldownloads.html has the SHAs of
those builds.

The standalone installers are made from the "full" packages, ie. contain more
packages than the bare minimum.


Jürgen


PS: the ongoing master build is for 8929e74a57, so it will have 1e715580af69.

-- 
Jürgen E. Fischer   norBIT GmbH Tel. +49-4931-918175-31
Dipl.-Inf. (FH) Rheinstraße 13  Fax. +49-4931-918175-50
Software Engineer   D-26506 Nordenhttps://www.norbit.de
QGIS release manager (PSC)  GermanyIRC: jef on FreeNode


signature.asc
Description: PGP signature
norBIT Gesellschaft fuer Unternehmensberatung und Informationssysteme mbH
Rheinstrasse 13, 26506 Norden
GF: Juergen Fischer, Nils Kutscher HR: Amtsgericht Aurich HRB 100827
Datenschutzerklaerung: https://www.norbit.de/83/
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] PG loading broken in master

[Sandro Santilli]

On Wed, Sep 16, 2020 at 10:38:28AM +0200, Andreas Neumann wrote:
> It seems like the second and third checkout from git bisect is now
> between the good old start and the bad commit. So apparently only the
> first checkout from git bisect was outside the range, the next one
> inside. Perhaps this has to do with the fact that the checkouts need to
> be dividable by a certain number?

Just a guess: merges with a parent _before_ the good commit were
present in the history from the bad commit ?

--strk;
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

[QGIS-Developer] Which qgis-dev in osgeo4w?

[Paolo Cavallini]

Hi all,
I see from
https://www.qgis.org/en/site/forusers/download.html
that the nightly is called qgis-dev-full
whereas in
http://download.osgeo.org/osgeo4w/x86_64/versions.html
it is called qgis-full-dev, but it does not seem up to date (2.99.0-2);
qgis-dev, 3.15.0-54 might be the correct one. Should the website be updated?
BTW, how can an user know which revision has been compiled, before
actually installing it?
Cheers.
-- 
Paolo Cavallini
www.faunalia.eu - QGIS.org
training, support, development on QGIS, PostGIS and more
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] PG loading broken in master

[Tim Sutton]

Glad to hear it!

On Wed, Sep 16, 2020 at 9:38 AM Andreas Neumann  wrote:

> It seems like the second and third checkout from git bisect is now between
> the good old start and the bad commit. So apparently only the first
> checkout from git bisect was outside the range, the next one inside.
> Perhaps this has to do with the fact that the checkouts need to be
> dividable by a certain number?
>
> Anyway - I'm getting closer to the culprit ;-)
>
> Andreas
>
> On 2020-09-16 10:28, Tim Sutton wrote:
>
> Hi
>
> Hmm I never had that problem - I just googled for a tutorial for you
> though so I didnt test to see if the tutorial was doing things correctly -
> you may want to try another source of info e.g.
>
> https://git-scm.com/docs/git-bisect
>
> Regards
>
> Tim
>
> On Wed, Sep 16, 2020 at 8:40 AM Andreas Neumann 
> wrote:
>
> Ok - I am trying to learn git bisect, but I'm really confused to how it
> works:
>
> Here is what I did:
>
> $ git bisect start
>
> $ git bisect good c6e28828dd3c9d0459033643ced4ba8994d7bc92
>
> $ git bisect bad 1c2b4ae2b0f3add7767d473f2b4137d7de9afd24
> Bisecting: 18 revisions left to test after this (roughly 4 steps)
> [89e4d4925a7ff26561acde84a861986cb30af961] enable QgsMeshTriangulation
> with SIP and change SIP_CONCAT_PARTS
>
> Why on earth is git bisect checking out something older than the good
> commit? Really strange ...
>
> I would have expected that git bisect would check out something between
> the good and the bad commit.
>
> Confused ...
>
> Andreas
>
>
>
> On 2020-09-15 23:35, Tim Sutton wrote:
>
> Hi Andreas
>
>
> If you are not already aware of it, check out git bisect, which helps you
> to isolate a breaking commit quickly.
>
>
> https://www.metaltoad.com/blog/beginners-guide-git-bisect-process-elimination
>
> Regards
>
> Tim
>
> On Tue, 15 Sep 2020, 16:46 Andreas Neumann,  wrote:
>
> Hi Jorge,
>
> There is only one authcfg. The service configuration doesn't contain auth
> information - just host, port and db. It is quite normal that you have a
> service name and an authcfg in the same data source.
>
> I am currently compiling older commits to find out which commit still
> worked.
>
> Probably I know more tomorrow.
>
> Greetings,
>
> Andreas
>
> On 2020-09-15 17:40, Jorge Gustavo Rocha wrote:
>
> Hi Andreas,
>
> Just a quick question: if both are present, should one be ignored? Which
> auth configuration should be used in that case?
>
> Regards,
>
> Jorge
>
> On 15/09/20 16:05, Andreas Neumann wrote:
>
> Ok -
>
> I am currently building older versions of master to find out which
> commit still worked.
>
> Will let you know, if I know more.
>
> Thanks,
>
> Andreas
>
> On 2020-09-15 17:02, Alessandro Pasotti wrote:
>
> On Tue, Sep 15, 2020 at 4:52 PM Andreas Neumann  > wrote:
>
>
> Hi,
>
> Some recent commit broke all my projects that load PG layers with a
> datasource like:
>
> service='pub' authcfg=sogis00 key='t_id' srid=2056
> type=Point checkPrimaryKeyUnicity='1' table="schema"."table"
> (geometrie) sql=
>
> Means a combination of PG service name and auth config. None of the
> PG layers in the project can be loaded.
>
> Any idea what recent commit might be the culprit?
>
>
>
> Sorry I don't, but I'm in bug fixing mode right now and working on
> other PG issues, if you have any chance to let me debug your project
> please let me know.
>
> Cheers
>
>
>
>
> ___
> QGIS-Developer mailing list
> QGIS-Developer@lists.osgeo.org
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>
>
> J. Gustavo
>
>
> ___
> QGIS-Developer mailing list
> QGIS-Developer@lists.osgeo.org
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>
>
>
>
> --
>
> --
> ​
> Tim Sutton
> Visit http://kartoza.com to find out about open source:
>  * Desktop GIS programming services
>  * Geospatial web development
> * GIS Training
> * Consulting Services
> Skype: timlinux Irc: timlinux on #qgis at freenode.net
> Tim is a member of the QGIS Project Steering Committee
>
> ---
> Kartoza is a merger between Linfiniti and Afrispatial
>
>
>

-- 
--
​

Tim Sutton
Visit http://kartoza.com to find out about open source:
 * Desktop GIS programming services
 * Geospatial web development
* GIS Training
* Consulting Services
Skype: timlinux Irc: timlinux on #qgis at freenode.net
Tim is a member of the QGIS Project Steering Committee
---
Kartoza is a merger between Linfiniti and Afrispatial

Re: [QGIS-Developer] PG loading broken in master

[Andreas Neumann]

It seems like the second and third checkout from git bisect is now
between the good old start and the bad commit. So apparently only the
first checkout from git bisect was outside the range, the next one
inside. Perhaps this has to do with the fact that the checkouts need to
be dividable by a certain number? 

Anyway - I'm getting closer to the culprit ;-) 

Andreas 


On 2020-09-16 10:28, Tim Sutton wrote:

Hi 

Hmm I never had that problem - I just googled for a tutorial for you though so I didnt test to see if the tutorial was doing things correctly - you may want to try another source of info e.g. 

https://git-scm.com/docs/git-bisect 

Regards 

Tim 

On Wed, Sep 16, 2020 at 8:40 AM Andreas Neumann  wrote: 

Ok - I am trying to learn git bisect, but I'm really confused to how it works: 

Here is what I did: 

$ git bisect start
$ git bisect good c6e28828dd3c9d0459033643ced4ba8994d7bc92  
$ git bisect bad 1c2b4ae2b0f3add7767d473f2b4137d7de9afd24 
Bisecting: 18 revisions left to test after this (roughly 4 steps) 
[89e4d4925a7ff26561acde84a861986cb30af961] enable QgsMeshTriangulation with SIP and change SIP_CONCAT_PARTS 

Why on earth is git bisect checking out something older than the good commit? Really strange ... 

I would have expected that git bisect would check out something between the good and the bad commit. 

Confused ... 

Andreas   

On 2020-09-15 23:35, Tim Sutton wrote: 

Hi Andreas 

If you are not already aware of it, check out git bisect, which helps you to isolate a breaking commit quickly. 

https://www.metaltoad.com/blog/beginners-guide-git-bisect-process-elimination 

Regards 

Tim 

On Tue, 15 Sep 2020, 16:46 Andreas Neumann,  wrote: 

Hi Jorge, 

There is only one authcfg. The service configuration doesn't contain auth information - just host, port and db. It is quite normal that you have a service name and an authcfg in the same data source. 

I am currently compiling older commits to find out which commit still worked. 

Probably I know more tomorrow. 

Greetings, 

Andreas 

On 2020-09-15 17:40, Jorge Gustavo Rocha wrote: 
Hi Andreas,


Just a quick question: if both are present, should one be ignored? Which
auth configuration should be used in that case?

Regards,

Jorge

On 15/09/20 16:05, Andreas Neumann wrote: Ok - 


I am currently building older versions of master to find out which
commit still worked.

Will let you know, if I know more.

Thanks,

Andreas

On 2020-09-15 17:02, Alessandro Pasotti wrote:

On Tue, Sep 15, 2020 at 4:52 PM Andreas Neumann > wrote: 
Hi,


Some recent commit broke all my projects that load PG layers with a
datasource like:

service='pub' authcfg=sogis00 key='t_id' srid=2056
type=Point checkPrimaryKeyUnicity='1' table="schema"."table"
(geometrie) sql=

Means a combination of PG service name and auth config. None of the
PG layers in the project can be loaded.

Any idea what recent commit might be the culprit? 


Sorry I don't, but I'm in bug fixing mode right now and working on
other PG issues, if you have any chance to let me debug your project
please let me know.

Cheers


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

J. Gustavo 


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer 


--

--


​ 
Tim Sutton 
Visit http://kartoza.com [1] to find out about open source:

* Desktop GIS programming services
* Geospatial web development 
* GIS Training 
* Consulting Services 

Skype: timlinux Irc: timlinux on #qgis at freenode.net [2] 

Tim is a member of the QGIS Project Steering Committee 
---


Kartoza is a merger between Linfiniti and Afrispatial 




Links:
--
[1] http://kartoza.com/
[2] http://freenode.net/___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] PG loading broken in master

[Tim Sutton]

Hi

Hmm I never had that problem - I just googled for a tutorial for you though
so I didnt test to see if the tutorial was doing things correctly - you may
want to try another source of info e.g.

https://git-scm.com/docs/git-bisect

Regards

Tim

On Wed, Sep 16, 2020 at 8:40 AM Andreas Neumann  wrote:

> Ok - I am trying to learn git bisect, but I'm really confused to how it
> works:
>
> Here is what I did:
>
> $ git bisect start
>
> $ git bisect good c6e28828dd3c9d0459033643ced4ba8994d7bc92
>
> $ git bisect bad 1c2b4ae2b0f3add7767d473f2b4137d7de9afd24
> Bisecting: 18 revisions left to test after this (roughly 4 steps)
> [89e4d4925a7ff26561acde84a861986cb30af961] enable QgsMeshTriangulation
> with SIP and change SIP_CONCAT_PARTS
>
> Why on earth is git bisect checking out something older than the good
> commit? Really strange ...
>
> I would have expected that git bisect would check out something between
> the good and the bad commit.
>
> Confused ...
>
> Andreas
>
>
>
> On 2020-09-15 23:35, Tim Sutton wrote:
>
> Hi Andreas
>
>
> If you are not already aware of it, check out git bisect, which helps you
> to isolate a breaking commit quickly.
>
>
> https://www.metaltoad.com/blog/beginners-guide-git-bisect-process-elimination
>
> Regards
>
> Tim
>
> On Tue, 15 Sep 2020, 16:46 Andreas Neumann,  wrote:
>
> Hi Jorge,
>
> There is only one authcfg. The service configuration doesn't contain auth
> information - just host, port and db. It is quite normal that you have a
> service name and an authcfg in the same data source.
>
> I am currently compiling older commits to find out which commit still
> worked.
>
> Probably I know more tomorrow.
>
> Greetings,
>
> Andreas
>
> On 2020-09-15 17:40, Jorge Gustavo Rocha wrote:
>
> Hi Andreas,
>
> Just a quick question: if both are present, should one be ignored? Which
> auth configuration should be used in that case?
>
> Regards,
>
> Jorge
>
> On 15/09/20 16:05, Andreas Neumann wrote:
>
> Ok -
>
> I am currently building older versions of master to find out which
> commit still worked.
>
> Will let you know, if I know more.
>
> Thanks,
>
> Andreas
>
> On 2020-09-15 17:02, Alessandro Pasotti wrote:
>
> On Tue, Sep 15, 2020 at 4:52 PM Andreas Neumann  > wrote:
>
>
> Hi,
>
> Some recent commit broke all my projects that load PG layers with a
> datasource like:
>
> service='pub' authcfg=sogis00 key='t_id' srid=2056
> type=Point checkPrimaryKeyUnicity='1' table="schema"."table"
> (geometrie) sql=
>
> Means a combination of PG service name and auth config. None of the
> PG layers in the project can be loaded.
>
> Any idea what recent commit might be the culprit?
>
>
>
> Sorry I don't, but I'm in bug fixing mode right now and working on
> other PG issues, if you have any chance to let me debug your project
> please let me know.
>
> Cheers
>
>
>
>
> ___
> QGIS-Developer mailing list
> QGIS-Developer@lists.osgeo.org
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>
>
> J. Gustavo
>
>
> ___
> QGIS-Developer mailing list
> QGIS-Developer@lists.osgeo.org
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>
>
>

-- 
--
​

Tim Sutton
Visit http://kartoza.com to find out about open source:
 * Desktop GIS programming services
 * Geospatial web development
* GIS Training
* Consulting Services
Skype: timlinux Irc: timlinux on #qgis at freenode.net
Tim is a member of the QGIS Project Steering Committee
---
Kartoza is a merger between Linfiniti and Afrispatial
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] PG loading broken in master

[Andreas Neumann]

Ok - I am trying to learn git bisect, but I'm really confused to how it
works: 

Here is what I did: 

$ git bisect start  
   
   
  
$ git bisect good c6e28828dd3c9d0459033643ced4ba8994d7bc92  
   
 
$ git bisect bad 1c2b4ae2b0f3add7767d473f2b4137d7de9afd24 
Bisecting: 18 revisions left to test after this (roughly 4 steps) 
[89e4d4925a7ff26561acde84a861986cb30af961] enable QgsMeshTriangulation
with SIP and change SIP_CONCAT_PARTS 


Why on earth is git bisect checking out something older than the good
commit? Really strange ... 


I would have expected that git bisect would check out something between
the good and the bad commit. 

Confused ... 

Andreas 
   
   


On 2020-09-15 23:35, Tim Sutton wrote:

Hi Andreas 

If you are not already aware of it, check out git bisect, which helps you to isolate a breaking commit quickly. 

https://www.metaltoad.com/blog/beginners-guide-git-bisect-process-elimination 

Regards 

Tim 

On Tue, 15 Sep 2020, 16:46 Andreas Neumann,  wrote: 

Hi Jorge, 

There is only one authcfg. The service configuration doesn't contain auth information - just host, port and db. It is quite normal that you have a service name and an authcfg in the same data source. 

I am currently compiling older commits to find out which commit still worked. 

Probably I know more tomorrow. 

Greetings, 

Andreas 

On 2020-09-15 17:40, Jorge Gustavo Rocha wrote: 
Hi Andreas,


Just a quick question: if both are present, should one be ignored? Which
auth configuration should be used in that case?

Regards,

Jorge

On 15/09/20 16:05, Andreas Neumann wrote: Ok - 


I am currently building older versions of master to find out which
commit still worked.

Will let you know, if I know more.

Thanks,

Andreas

On 2020-09-15 17:02, Alessandro Pasotti wrote:

On Tue, Sep 15, 2020 at 4:52 PM Andreas Neumann > wrote: 
Hi,


Some recent commit broke all my projects that load PG layers with a
datasource like:

service='pub' authcfg=sogis00 key='t_id' srid=2056
type=Point checkPrimaryKeyUnicity='1' table="schema"."table"
(geometrie) sql=

Means a combination of PG service name and auth config. None of the
PG layers in the project can be loaded.

Any idea what recent commit might be the culprit? 


Sorry I don't, but I'm in bug fixing mode right now and working on
other PG issues, if you have any chance to let me debug your project
please let me know.

Cheers


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

J. Gustavo 


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer