thanks Tim - I'm cc'ing Matthew Brett here since I'm not sure if he's on
this list.
cheers
russ



On Sun, Mar 2, 2014 at 3:13 PM, Timothy Coalson <[email protected]> wrote:

> From the .dtd on https://www.nitrc.org/projects/gifti/, I see this:
>
>       Endian       (BigEndian |
>                     LittleEndian)  #REQUIRED
>
> From GIFTI_Surface_Format.pdf on the same site (which also echos the above
> on page 7), it looks like line breaks are also implicitly forbidden:
>
> "The second encoding, Base64Binary (http://www.ietf.org/rfc/rfc3548.txt or
> http://email.about.com/cs/standards/a/base64_encoding.htm), encodes binary
> data into a sequence of ASCII characters."
>
> From http://www.ietf.org/rfc/rfc3548.txt:
>
>    Implementations MUST NOT not add line feeds to base encoded data
>    unless the specification referring to this document explicitly
>    directs base encoders to add line feeds after a specific number of
>    characters.
>
> I think nibabel needs to change how it writes gifti files.
>
> Tim
>
>
>
> On Sat, Mar 1, 2014 at 3:13 PM, Russ Poldrack <[email protected]> wrote:
>
>> hi all - it turns out that gifti files created by nibabel.gifti.giftiio
>> are not readable by connectome workbench, due to two main problems.  first,
>>  the files generated by nibabel use the endian code 'GIFTI_ENDIAN_LITTLE'
>> whereas workbench seems to require the alternate code 'LittleEndian'.
>>  second, nibabel inserts line breaks with the <Data> section of the file,
>> but workbench doesn't seem to be able to handle those.  I don't know the
>> gifti standard well enough to know which of these is breaking the standard,
>> but it would be great if the workbench could handle the files generated by
>> nibabel. in the meantime I have generated a hacky little script that can
>> fix the files generated by nibabel to make them compatible with the
>> workbench:
>>
>> https://github.com/poldrack/wbench/blob/master/wbify_gifti.py
>>
>> cheers
>> russ
>>
>>  _______________________________________________
>> HCP-Users mailing list
>> [email protected]
>> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>>
>
>

_______________________________________________
HCP-Users mailing list
[email protected]
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

Reply via email to