[Live-demo] OSGeo-Live 8.0beta1testing feedback
Even Rouault
even.rouault at spatialys.com
Sat Aug 9 10:09:31 PDT 2014
Hi,
here's my experience with my more or less random testing the osgeo-live-
mini-8.0beta1-i386.iso. A few issues for which I have created tickets, but
other topics which are more subject to discussion.
Overall you've done a great job !
Tested with VirtualBox 4.2.12 with 1GB RAM allocated to the VM.
Live launched in French.
Translation issues :
- The welcome message is non translated
- The main folders are non translated on the Desktop or in the menu : "Browser
Clients", "Crisis Management", "Help" etc...
- When launching the Help it starts in English and not on /fr/index.html
- In the help, the top menu is untranslated "Home", "Contents", "Download" ...
Help issues :
- QGIS : Software version is announced 1.8.0 in the French translation. And I
noticed the English version is also behind : 2.0.1. --> Perhaps software
version numbers should not be announced at all, to avoid being always behind ?
Or that should be somehow automated ?
I have opened the blue marble raster dataset in QGIS. Then when opening
~/data/vector/world_merc.shp in QGIS, it didn't overlay on top of it. The
reason is that QGIS doesn't recognize the SRS from the .prj file. Well not a
OSGeo Live issue by it self, but more a OGR / QGIS one ;-) The strange thing
is that when clicking on the Specify button in the Properties dialog box of
the layer it correctly proposes a "reasonable" PROJ.4 string (not the right
one that corresponds to EPSG:3857, but that's expected from what GDAL
understand from the .prj), but didn't select it at the beginning (QGIS issue).
Once validated, the vector reasonably overlays on top of the raster.
There's a menu Geospatial --> Navigation And Maps --> OpenStreet Map, with 3
items in it : JOSM , Merkaator, and View OSM online. Those 3 items are present
in the upper Navigation And Maps menu. What's the point of the OpenStreet Map
submenu ?
Same issue with Browser Clients and the Cartaro, Geomajas, GeoNode, MapFish
submenus. Actually, on reflection, it would probably better to keep the
submenus, but remove the duplicate entries from the toplevel menu ?
The main menu has an Education submenu which lists part of the geospatial
software, without any obvious logic. Perhaps it could be just be deleted ? The
Geospatial submenu is much more organized.
When clicking on View OSM online, either in " Navigation And Maps" menu or its
"OpenStreet Map", nothing happens if the browser is not already started. If it
is started, OK.
Similar issue with the MapServer demo (Webservices/Mapserver), ZOO Project
Demo, QGIS Server, etc...
Hum: actually after restarting the LiveDVD, I have not that issue anymore... I
don't know what I did the first time to go into that situation.
When having launched Firefox and then the ZOO Project Demo, it seems broken :
a pink rectangle is displayed. When panning one can set a world map for a few
moments, but it is instantenously replaced by the pink rectangle once the
mouse button is released.
Hum; actually after restarting the LiveDVD, I have not that issue anymore...
When having launched Firefox and then the QGIS_MapServer demo, panning works.
But once you zoom in, blank map or partly blank with partly valid areas.
Likely a QGIS_MapServer issue ?
Launching afterwards the pycsw demo crashed Firefox and the whole Virtualbox
VM hanged (the mouse cursor didn't move any more). I had to stop it. After
reboot, the demo started.
Hum, I have had another VM hang a bit after running again the QGIS_MapServer
demo and zooming in/out very quicly. A huge number of qgis_mapserver processes
are forked. A top showed the load to go up to 50. This is actually quite
easily reproducable (each time I tried that demo). So pycsw is fine. It is just
I launched it while qgis_mapserver was beginning to bring the VM down.
--> Filed as https://trac.osgeo.org/osgeo/ticket/1383
When browsing on the "Sample data", for example in natural_earth2, if I click
on the HYP_50M_SR_W.tif file, the default image viewer opens. Which could be
inappropriate for very big tiff. Wouldn't it be more appropriate to associate
.tif with QGIS and launch QGIS ? Update: I see it is proposed on the
contextual menu with a right click. And actually it only proposes QGIS whereas
other Desktop GIS could open it : OpenJump, etc.
But changing the default would still be nice. For example
north_carolina/rast_geotiff/elevid_D792_6m.tif cannot be opened by the default
image browser.
The GeoMOOSE demo launches but nothing displays. It gets stucked on trying to
connect to api.maps.yahoo.com (which pings fine from a console in the VM)
--> Filed as https://trac.osgeo.org/osgeo/ticket/1384
Start Geomajas: a browser opens with a HTTP ERROR : 503. Problem accessing
/showcase. Service Unavailable
--> Filed as https://trac.osgeo.org/osgeo/ticket/1385
QGIS: not a big deal, but when opening both .qgs in qgis-samples (itasca and
naturalearth), a warning is displayed about the .qgs not being in the latest
version.
~/data/kml: the directory is empty
Suggestion: ~/data/natural_earth2/HYP_50M_SR_W.tif weighs 175 MB. If you are
short of DVD space, "gdal_translate HYP_50M_SR_W.tif out.tif -co compress=jpeg
-co photometric=ycbcr -co tiled=yes" produces a 3.75 MB file... (deflate
compression, which is lossless, brings to ~ 80 MB, but I guess the files in the
DVD itselves must be compressed with zlib, so that would be bring no
advantage). Of course that could potentially cause some incompatibilities with
applications that have poor support for some formulation of GeoTIFF. But at
least would work fine with all GDAL-based stack : QGIS etc...
Trying to cleanly stop the VM doesn't kill the VirtualBox session. The last
message displayed is "Stopping early crypto disks...". And it actually
"ejects" the .iso from the VM configuration. So next time I start the VM, I
have to reattach the .iso.
Unrelated remark : not a big issue, wouldn't it be more meaningful to have a
dedicated Trac instance for osgeo-live rather than the generic osgeo one ?
That way you could use the Component submenu to list the applications, etc.
But I can understand that it would mean more work for the folks at SAC and
there would be a need for a migration from the existing bugs. Probably not
worth it.
Regards,
Even
--
Spatialys - Geospatial professional services
http://www.spatialys.com
More information about the Osgeolive
mailing list