Hi Andreas, Steve and others, On Mon, Aug 09, 2004 at 10:44:08PM +0200, Andreas Tille wrote: > On Mon, 9 Aug 2004, Steve Halasz wrote: > > > ... > > > > * Package name : grass5.7 > > > > Version : 5.7.0 > > > Do you think that we need two different versions of > > > GRASS in Debian? > > > > The motivation for this is bug #248649 [1]. I package > > qgis which has a plugin for reading and editing grass > > layers. This plugin depends on gdal support for grass > > written by Radim Blazek. I have verified with Radim > > that his patches will only work with grass 5.7's newly > > provided libraries. > > > > Also, I believe 5.7 fixes bug #234275 [2]. I am not a > > heavy grass user, but it is my impression that 5.0 is > > getting old and that 5.7 is greatly improved. The > > roadmap [3] for grass is kind of confusing, but it > > seems reasonable to package 5.0.x, 5.3, and 5.7 to > > provide various levels of stability vs. currentness. > So far we have at least two reasons to package GRASS 5.7. My > question was not against your plan to package 5.7 but instead just > to upgrade the existing GRASS. I'm not a GRASS user nor a GIS > expert (that's why I CCed Silke) but IMHO a packaging team which > would target to a recent GRASS version would be the best solution > here. I'd like to explain how the different grass versions are connected to each other: Historically grass has been a more raster oriented GIS than a vector GIS. This is the state of grass 5.0 which has very good opportunities to work with raster data and a rather old fashioned vector modell so that it is not much fun to work with the vector stuff. This is the reason why Radim wrote a new vector model which is now in a rather stable state. grass 5.7 is the version which contains this vector model. As far as I understood this will be version 6.0 as soon as the grass developers decided that this is really stable (they have also a very strong definition of 'stable' similar as the debian folks). The raster model is almost the same as in grass 5.0 but has been revised a lot so that duplicated code has been removed. grass 5.3 is nearly the same as grass 5.0 (i.e. without the new vector modell) but contains some more features for datum transformation and a lot of bug fixes that have not been backported to grass 5.0 and will not be backported. > > > I sent a message to the grass packager (Federico Di > > Gregorio <fog@debian.org>) about a week ago asking > > about all this and have received no response. There is > > also a wishlist bug for grass 5.3 (261483 [4]) that > > has received no response. I hope that he will have had > > sufficient time to raise objections before I get the > > package squared away. > Once I had some conversation with Federico about GRASS and if > I remember right he would not have a problem with comaintenance > because of time constraints - but I might remember this false > or the situation might have changed. > > So if I would be you I would proceed as follows: > > - Change the ITP bug into a "new version available" bug against GRASS. > - Provide a patch which build a working GRASS package of version 5.7. Regarding the information that I gave about the different version I would second this proposition. Another possibilty could be to update grass 5.0 to grass 5.3 and to add a new package grass 5.7 but since the raster stuff of grass 5.[0|3] is nearly the same as for grass 5.7 this is not really necessary. So far, Silke -- Silke Reimer Intevation GmbH http://intevation.de/ FreeGIS http://freegis.org/
Attachment:
pgpQPZyQUw5Km.pgp
Description: PGP signature