Opened 12 years ago

Closed 12 years ago

#3922 closed defect (fixed)

Edge cases in NITFCreate don't conform to MIL-STD-2500-C

Reported by: klaricmn Owned by: warmerdam
Priority: normal Milestone: 1.9.0
Component: GDAL_Raster Version: svn-trunk
Severity: normal Keywords: nitf


I believe that I have found a few edge cases when the NITFCreate method does not conform to the NITF MIL-STD-2500-C standard. The cases that I'm describing can be found in paragraph (page 41 of the PDF file).

The existing code relevant to this paragraph was modified for #3263, but only handles two of the four cases. It handles the case when the block size is large (i.e. >8192) in both dimensions and when it has small block sizes in both dimensions. However, it does not currently account for the case when only one of the block dimensions exceeds 8192 even though the spec allows for it.

The attached patch addresses this issue.

Attachments (1)

nitf-big-block.patch (975 bytes ) - added by klaricmn 12 years ago.

Download all attachments as: .zip

Change History (3)

by klaricmn, 12 years ago

Attachment: nitf-big-block.patch added

comment:1 by Even Rouault, 12 years ago

Component: defaultGDAL_Raster
Milestone: 1.8.0

Yes those 2 cases were unadressed. I've modified a bit your patch since it didn't initialize the nNBPC or nNBPR variables.

r21542 /trunk/ (autotest/gdrivers/ gdal/frmts/nitf/nitffile.c): NITFCreate(): deal with cases where image_height = block_height > 8192 or image_width = block_width > 8192 (#3922)

comment:2 by Even Rouault, 12 years ago

Milestone: 1.9.0
Resolution: fixed
Status: newclosed
Note: See TracTickets for help on using tickets.