[GRASS-dev] Re: [GRASS GIS] #1006: r.terraflow fails to stat() stream file on Windows

[GRASS GIS]

#1006: r.terraflow fails to stat() stream file on Windows
-----------------------------------+----------------------------------------
 Reporter:  marisn                 |       Owner:  grass-dev@…              
     Type:  defect                 |      Status:  new                      
 Priority:  critical               |   Milestone:  6.4.0                    
Component:  Raster                 |     Version:  svn-releasebranch64      
 Keywords:  wingrass, r.terraflow  |    Platform:  MSWindows Vista          
      Cpu:  x86-32                 |  
-----------------------------------+----------------------------------------

Comment(by mmetz):

 Replying to [comment:36 hamish]:
 > Replying to [comment:35 mmetz]:
 > > the ability to process large datasets is also covered by other
 modules.
 >
 > I know that, but the (perhaps recently historic) way the modules are
 presented in the menus does not make that clear to users.

 Well, then we have to change the way modules are presented in the menus.
 >
 > from raster/r.terraflow/filldepr.cc
 > {{{
 >     /*read next edge*/
 >     ae = boundaryStr->read_item(&nextedge);
 >     assert(ae == AMI_ERROR_NO_ERROR);
 > }}}
 >
 > which calls include/iostream/mem_stream.h
 > {{{
 > template<class T>
 > AMI_err MEM_STREAM<T>::read_item(T **elt)  {
 >
 >   assert(data);
 >
 >   if(curr == dataend) {
 >     return AMI_ERROR_END_OF_STREAM;
 >   }
 >   *elt = curr;
 >   curr++;
 >   return AMI_ERROR_NO_ERROR;
 > };
 > }}}
 >
 > so it seems that it is prematurely reaching the end of the stream?

 I never thought I would agree to fix depression filling... The mingw32
 version of stream_len() does not work 100%, workaround in r42734 and
 r42735. Worksforme now on Windows. On Linux, no difference in results. Can
 I now put warning messages in r.terraflow that its hydrological algorithms
 from the late 80ies and early 90ies do not produce very accurate results
 (its emphasis is on the computational complexity of the algorithms, rather
 than on modeling realistic flow)?

 > > r.terraflow and iostream is written for Linux, as also previous
 tickets have
 > > shown, no reason to expect that this is the last problem with Windows.
 > > Please leave it for 6.5 or 7.0.
 >
 > we could easily say the same for all of GRASS.

 Most parts of grass are not giving problems with mingw32, maybe because
 these adhere to C89? C++ doesn't, so problems might be expected here.

 > [snip] even if it is crippled due to LFS (you might say 32bit XP is too
 with its 3.2gb RAM limit),

 You might say that mingw32 is the problem (apart from the msys problems).
 IIUC, stuff compiled on e.g. Windows7 64bit with mingw32 is still 32bit
 and has all the 32bit limits, even if the OS is 64bit. Plus other
 annoyances like sizeof(long) = 4 on Windows 64bit.

 >
 > in pursuit of a technical fix I will add debug messages to mem_stream.h.
 (r42731)

 IMHO, the debug messages can be removed now, then the fix can be
 backported to 6.4.1 ;-)

 Markus M

-- 
Ticket URL: <https://trac.osgeo.org/grass/ticket/1006#comment:37>
GRASS GIS <http://grass.osgeo.org>