<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Very well. I am glad we are at the same boat here ;-)</p>
    <p>Andreas<br>
    </p>
    <div class="moz-cite-prefix">Am 08.05.20 um 12:43 schrieb Matthias
      Kuhn:<br>
    </div>
    <blockquote type="cite"
      cite="mid:6e7e9c88-af43-2ea0-c364-83a6ac3ed376@opengis.ch">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <p>Hi Andreas</p>
      <p>Thanks for also pointing out this question. I would like to
        completely focus on b) for now to not disperse the discussion
        too much.</p>
      <p>If it comes down to unresolvable problems we can still go for
        d).</p>
      <p>But let's keep that for later.</p>
      <p>Matthias<br>
      </p>
      <div class="moz-cite-prefix">On 5/8/20 11:59 AM, Andreas Neumann
        wrote:<br>
      </div>
      <blockquote type="cite"
        cite="mid:77fa2b0a-be80-7f65-2d64-f267c2930962@carto.net">
        <meta http-equiv="Content-Type" content="text/html;
          charset=UTF-8">
        <p>Hi Matthias,</p>
        <p>Thank you for listing all the open issues/problem with gpkg
          that you know of. It really helps.</p>
        <p>If we don't like gpkg as default format - the question is:
          what is the alternative?</p>
        <p>a) stay with ESRI shapefile (I think noone would like this)<br>
          b) work with the SQLite, gpkg/OGC community to fix the gpkg
          issues (my preference)<br>
          c) use ESRI FGDB format (then we are at the mercy of ESRI)<br>
          d) invent something new (risky, if only QGIS uses that,
          interoperability would suck)</p>
        <p>I would prefer option b) a lot, and if that is not feasible,
          then maybe d). d) will also be risky.<br>
        </p>
        <p>a) would equally suck as the current state of gpkg - I've
          seen far too many corrupt shape files, people complaining
          about interoperability issues (ArcGIS would show features that
          had been deleted in QGIS, ) and I don't need to repeat the
          list of the numerous restrictions of ESRI shp format.</p>
        <p>Andreas<br>
        </p>
        <div class="moz-cite-prefix">Am 08.05.20 um 11:30 schrieb
          Matthias Kuhn:<br>
        </div>
        <blockquote type="cite"
          cite="mid:4432048f-efef-5d9c-51cc-7a50af2ab443@opengis.ch">
          <meta http-equiv="content-type" content="text/html;
            charset=UTF-8">
          <p>Hi list,</p>
          <p>I wondered about the state of GeoPackage. Personally, cince
            it has been introduced to qgis and evenmore since it has
            been selected as the default format, I have never grown to
            fully and completely.</p>
          <p>I do not want to trigger a evangelical discussion here. I'd
            like to see where we are and what we can reasonably do to
            have a default file format which can be recommended with no
            bad feelings.<br>
          </p>
          <p><br>
          </p>
          <p>Here follow a couple of observations over the years, some
            of them properties of the specs I believe:</p>
          <p><br>
          </p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">* The fid requirement</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  I sometimes want my features to be
              identified by uuids or others. They also tend to
              accumulate if derived datasets are created (through
              processing etc). If I need some pseudo stable primary key
              there is a rowid builtin into sqlite, we don't need a
              second one.<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Possible mitigation: alter the ogr
              implementation. possibly alter the standard (required?)<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"></span><span id=":ws.co" class="tL8wMe
              EMoHub" style="text-align: left;" dir="ltr"><span
                id=":ws.co" class="tL8wMe EMoHub" style="text-align:
                left;" dir="ltr">* </span>The modification on r/o open</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Has caused too much pain on git.</span></p>
          <p style=" margin-top:0px; margin-bottom:0px; margin-left:0px;
            margin-right:0px; -qt-block-indent:0; text-indent:0px;
            -qt-user-state:0;"><span id=":ws.co" class="tL8wMe EMoHub"
              style="text-align: left;" dir="ltr">  Possible mitigation:
              a) switch to journal mode=delete (not an easy option
              because of </span><a class="moz-txt-link-freetext"
              href="https://issues.qgis.org/issues/15351"
              moz-do-not-send="true">https://issues.qgis.org/issues/15351</a>)
            b) only switch to wal mode when layers are put into edit
            mode (I have strong doubts this is a safe thing to do)<br>
          </p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"></span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"></span><span id=":ws.co" class="tL8wMe
              EMoHub" style="text-align: left;" dir="ltr"><span
                id=":ws.co" class="tL8wMe EMoHub" style="text-align:
                left;" dir="ltr">* </span>The network share freeze</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Our default file should play nicely
              with (windows) network shares. It's clear to everyone that
              we can't expect concurrent writes. But it should "just
              work" for concurrent read by many.</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Possible mitigation: switch to journal
              mode=delete for network shares (we are looking into this)<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"></span><span id=":ws.co" class="tL8wMe
              EMoHub" style="text-align: left;" dir="ltr"><span
                id=":ws.co" class="tL8wMe EMoHub" style="text-align:
                left;" dir="ltr">* </span>The wal file appearing next
              to the file</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  It is confusing to newcomers and looks
              almost like a sidecar file. I would care less if it was
              put into some system cache folder instead of just into my
              data folder. Or at least if it was a hidden file.<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Possible mitigation: switch to journal
              mode=delete (</span><span id=":ws.co" class="tL8wMe
              EMoHub" style="text-align: left;" dir="ltr"><span
                id=":ws.co" class="tL8wMe EMoHub" style="text-align:
                left;" dir="ltr">not an easy option because of </span><a
                class="moz-txt-link-freetext"
                href="https://issues.qgis.org/issues/15351"
                moz-do-not-send="true">https://issues.qgis.org/issues/15351</a>)</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"></span><span id=":ws.co" class="tL8wMe
              EMoHub" style="text-align: left;" dir="ltr"><span
                id=":ws.co" class="tL8wMe EMoHub" style="text-align:
                left;" dir="ltr">* </span>The couple of corrupted files
              I have received over the years which could only be
              repaired by a command line "dump contents as sql and
              execute into new file"</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  I have not found a way to reproduce
              this. Some of them were produced by older qgis versions
              making it easy to violate foreign key constraints and hard
              to recover. This has been fixed.</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Possible mitigation: offer a "repair"
              option in qgis. Through processing or "on the fly" upon
              detection.<br>
            </span></p>
          <p>*<span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"><span id=":ws.co" class="tL8wMe EMoHub"
                style="text-align: left;" dir="ltr"> </span>Default
              value magic replace values on insert (with no possibility
              to pre-evaluate them)</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  E.g. a global sequence like on postgres
              would be nice. Can be worked around through default values
              in qgis though.<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Possible mitigation: a)add it as a
              feature to sqlite. b) use qgis default values. c) live
              with it.<br>
            </span></p>
          <p>*<span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"><span id=":ws.co" class="tL8wMe EMoHub"
                style="text-align: left;" dir="ltr"> </span>The
              requirement for a single geometry column per table</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  I just don't see a good reason to
              forbid that</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">  Possible mitigation: a) alter the
              standard. b) ignore the standard and patch the ogr
              implementation.<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"><br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">I wonder how others feel about these
              topics.</span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr"><br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">- Are there more pain points I forgot to
              list?<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">- Do you see more approaches to mitigate
              these problems?<br>
            </span></p>
          <p><span id=":ws.co" class="tL8wMe EMoHub" style="text-align:
              left;" dir="ltr">- Is someone already working on these
              issues?<br>
            </span></p>
          <p><br>
          </p>
          <p>It would be great to have a standard file format that we
            can fully trust. Let's make a reality check if GeoPackage
            can be this format.</p>
          <p>Best regards<br>
          </p>
          <div class="moz-signature">-- <br>
            <meta http-equiv="content-type" content="text/html;
              charset=UTF-8">
            <div class="moz-signature">
              <title></title>
              <div class="moz-signature"> <span style="text-align:
                  left; color: #000000; font-family: 'Verdana',
                  sans-serif; font-size: 10pt">Matthias Kuhn</span><br>
                <a href="mailto:matthias@opengis.ch" target="_blank"
                  moz-do-not-send="true"> <span style="text-align:
                    left; color: #000000; font-family: 'Verdana',
                    sans-serif; font-size: 8pt">matthias@opengis.ch</span>
                </a><br>
                <span style="text-align: left; color: #000000;
                  font-family: 'Verdana', sans-serif; font-size: 8pt"><a
                    href="tel:+41764356763" moz-do-not-send="true">+41
                    (0)76 435 67 63</a></span><br>
                <div> <a href="http://www.opengis.ch"
                    moz-do-not-send="true"> <img
                      moz-do-not-send="false"
                      src="cid:part5.0912AC31.956B1BD2@carto.net"
                      alt="OPENGIS.ch Logo" class="" width="200"
                      height="80"></a> </div>
              </div>
            </div>
          </div>
          <br>
          <fieldset class="mimeAttachmentHeader"></fieldset>
          <pre class="moz-quote-pre" wrap="">_______________________________________________
QGIS-Developer mailing list
<a class="moz-txt-link-abbreviated" href="mailto:QGIS-Developer@lists.osgeo.org" moz-do-not-send="true">QGIS-Developer@lists.osgeo.org</a>
List info: <a class="moz-txt-link-freetext" href="https://lists.osgeo.org/mailman/listinfo/qgis-developer" moz-do-not-send="true">https://lists.osgeo.org/mailman/listinfo/qgis-developer</a>
Unsubscribe: <a class="moz-txt-link-freetext" href="https://lists.osgeo.org/mailman/listinfo/qgis-developer" moz-do-not-send="true">https://lists.osgeo.org/mailman/listinfo/qgis-developer</a></pre>
        </blockquote>
        <br>
        <fieldset class="mimeAttachmentHeader"></fieldset>
        <pre class="moz-quote-pre" wrap="">_______________________________________________
QGIS-Developer mailing list
<a class="moz-txt-link-abbreviated" href="mailto:QGIS-Developer@lists.osgeo.org" moz-do-not-send="true">QGIS-Developer@lists.osgeo.org</a>
List info: <a class="moz-txt-link-freetext" href="https://lists.osgeo.org/mailman/listinfo/qgis-developer" moz-do-not-send="true">https://lists.osgeo.org/mailman/listinfo/qgis-developer</a>
Unsubscribe: <a class="moz-txt-link-freetext" href="https://lists.osgeo.org/mailman/listinfo/qgis-developer" moz-do-not-send="true">https://lists.osgeo.org/mailman/listinfo/qgis-developer</a></pre>
      </blockquote>
    </blockquote>
  </body>
</html>