[postgis-devel] PostGIS 2.1 "transaction control statements"

Paragon Corporation lr at pcorp.us
Sun Feb 10 01:24:35 PST 2013


Depends.  The files should not have been created in extensions folder of
your source tree which are the ones that are copied to extension folder of
PostgreSQL install.
 
But if you copied the generated postgis.sql from regular source folder then
yes that would explain it becuase the extension system would see that as the
file to use and then scream violently when it realized you tried to trick it
with an imposter :)
 
 

  _____  

From: postgis-devel-bounces at lists.osgeo.org
[mailto:postgis-devel-bounces at lists.osgeo.org] On Behalf Of Stephen Mather
Sent: Sunday, February 10, 2013 1:56 AM
To: PostGIS Development Discussion
Subject: Re: [postgis-devel] PostGIS 2.1 "transaction control statements"


But if I copied those scripts from my install into their correct final
location, that would explain the strange affect, eh? 

I think this is the result of some advanced work-arounds... .

Best,
Steve


On Sat, Feb 9, 2013 at 12:45 PM, Paragon Corporation <lr at pcorp.us> wrote:



If you compiled --without-raster then extensions should not have built at
all since they don't build without raster.
 
So that I suppose could be the problem.  if that is the case.  Please ticket
it.  The extension files shouldn't have even tried to build at all.
 
thanks,
Regina

  _____  


From: postgis-devel-bounces at lists.osgeo.org
[mailto:postgis-devel-bounces at lists.osgeo.org] On Behalf Of Stephen Mather

Sent: Friday, February 08, 2013 11:00 PM 

To: PostGIS Development Discussion
Subject: Re: [postgis-devel] PostGIS 2.1 "transaction control statements"


select version() || ' ' || ' ' || postgis_full_version();

yields

PostgreSQL 9.1.7 on x86_64-unknown-linux-gnu, compiled by gcc (Ubuntu/Linaro
4.7.2-2ubuntu1) 4.7.2, 64-bit  POSTGIS="2.1.0SVN r11081"
GEOS="3.4.0dev-CAPI-1.8.0 r3762" PROJ="Rel. 4.7.1, 23 September 2009"
LIBXML="2.8.0" LIBJSON="UNKNOWN"


Looking through my history, it was an SVN pull:

svn checkout http://svn.osgeo.org/postgis/trunk/ postgis-2.1.0SVN
cd postgis-2.1.0SVN/
./autogen.sh 
./configure --with-gui --with-topology
./configure --with-gui --without-raster --with-topology
make
make comments
make cheatsheets 
sudo make install
sudo ldconfig
sudo make comments-install


:D  Looks like I was feeling my way around in the dark a bit there... .



On Fri, Feb 8, 2013 at 10:20 PM, Paragon Corporation <lr at pcorp.us> wrote:



Which version of trunk are you running with and where did you get it from:
Direct from svn or tar ball (and if tar ball what link)?
 
I just tried with Winnie's 9.2 latest Windows build built fresh whenever
there is a change and it installed fine creating a new postgis. Didn't try
9.1 though.
 
 
select version() || ' ' || ' ' || postgis_full_version();
 
--
PostgreSQL 9.2.2, compiled by Visual C++ build 1600, 64-bit
POSTGIS="2.1.0SVN r11085" GEOS="3.4.0dev-CAPI-1.8.0 r0" PROJ="Rel. 4.8.0, 6
March 2012" GDAL="GDAL 1.9.2, released 2012/10/08" LIBXML="2.7.8"
LIBJSON="UNKNOWN" RASTER
 
 
 
If you are running an older -- its possible there was a temporary issue that
was fixed, though I'm not aware of one.  
In the make sed strips out all BEGIN COMMITS as well as a bunch of other
things that are illegal in extension scripts and replaces them with
equivalent extension commands.  It also adds additional helper functions it
uses during the process and then drops after.
 
 


  _____  


From: postgis-devel-bounces at lists.osgeo.org
[mailto:postgis-devel-bounces at lists.osgeo.org] On Behalf Of Stephen Mather

Sent: Friday, February 08, 2013 5:13 PM 

To: PostGIS Development Discussion
Subject: Re: [postgis-devel] PostGIS 2.1 "transaction control statements"


-- $Id: postgis.sql.in.c 10806 2012-12-06 17:56:37Z pramsey $
...
BEGIN;
...

Yes, the BEGIN and END are there.  So, how does it work?  In the MAKE of the
extension, sed is supposed to strip out all transaction statements, or this
just slipped into the SVN somehow?


Best,
Steve



On Fri, Feb 8, 2013 at 4:37 PM, Paragon Corporation <lr at pcorp.us> wrote:



Hmm that's the same version I'm runnning though I'm running mine under
mingw.
 
I thought maybe you were having a sed issue that someone else complained
about, though they were on Mac and had to do with sed compatibility switch.
 
In your extension folder of your PostgreSQL install, 
 
If you open up postgis-2.1.0SVN.sql
 
Does it start with BEGIN and end with a COMMIT;?
 
If it does -- remove those.  The postgis extension build line:
 
sql_bits/postgis.sql: ../../postgis/postgis.sql
 sed -e 's/BEGIN;//g' -e 's/COMMIT;//g'  $< > $@

 should have done this but it sounds like it didn't for some reason.  
 
 
Those commands aren't legal in an extension script since the extension
machinery already wraps an extension in a transaction.
 
Hope that helps,
Regina

  _____  


From: postgis-devel-bounces at lists.osgeo.org
[mailto:postgis-devel-bounces at lists.osgeo.org] On Behalf Of Stephen Mather

Sent: Friday, February 08, 2013 4:24 PM
To: PostGIS Development Discussion
Subject: Re: [postgis-devel] PostGIS 2.1 "transaction control statements"


GNU sed 4.2.1



On Fri, Feb 8, 2013 at 3:08 PM, Paragon Corporation <lr at pcorp.us> wrote:



It sounds like the BEGIN /END transaction did not get stripped in your
extension building.
 
What version of sed do you have?
 
Thanks,
Regina
http://www.postgis.us
http://postgis.net
 

  _____  

From: postgis-devel-bounces at lists.osgeo.org
[mailto:postgis-devel-bounces at lists.osgeo.org] On Behalf Of Stephen Mather
Sent: Friday, February 08, 2013 9:19 AM
To: postgis-devel at lists.osgeo.org
Subject: [postgis-devel] PostGIS 2.1 "transaction control statements"


Hi All,

       Bounce me to users, if appropriate, but I have put together an SVN
build of PostGIS on Ubuntu 12.10 and postgresql 9.1, with GEOS 3.4.0dev.
All went well in the configure/make/make install mantra.  When I got to the
CREATE EXTENSION step I got the following cryptic message:

"transaction control statements are not allowed with an extension script"


I've had no problem loading PostGIS using the old method with sql scripts,
but thought I'd highlight this oddity in case it was useful and as yet
unreported.

Best,

Steve Mather


_______________________________________________
postgis-devel mailing list
postgis-devel at lists.osgeo.org
http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-devel





_______________________________________________
postgis-devel mailing list
postgis-devel at lists.osgeo.org
http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-devel





_______________________________________________
postgis-devel mailing list
postgis-devel at lists.osgeo.org
http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-devel





_______________________________________________
postgis-devel mailing list
postgis-devel at lists.osgeo.org
http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-devel




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20130210/0d9a963b/attachment.html>


More information about the postgis-devel mailing list