25 Mar
2015
25 Mar
'15
10:49 p.m.
Hi all,
The new hashing code just uses the data from the first and last megabyte
and the size of each content file to compute its hash, rather than the MD5
of the whole file.
The only potential problem (I can think of) comes if you make a DCP, then
change the content file and re-make the DCP. If the hashes for the two
content files are the same, DCP-o-matic would re-use the video data from
the first content file.
It's unlikely enough, I think, that the massive performance increase is
worth it.
Best regards,
Carl
On Sun, 22 Mar 2015, Manuel AC via DCPomatic wrote:
I'm also curious, as this feature derives from my
request I was
worried that it will affect other users. But if I remember well, you
are now using different data to detect a modified file instead on
relying solely in the hash.
Manuel AC
On Sat, Mar 21, 2015 at 9:00 PM, David Lawrence via DCPomatic
<dcpomatic(a)carlh.net> wrote:
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
Use a
less accurate but much less time-consuming hashing method for content, dramatically
speeding up examination of large files.
On my Retina MacBook Pro, ingest of a 35-minute ProRes 4444 file went from ~10 minutes to
instantaneous. That’s one big improvement! Any downside to less accurate hashing?
—dhl
On Mar 20, 2015, at 4:17 PM, Carl Hetherington via DCPomatic <dcpomatic(a)carlh.net>
wrote:
> Hi all,
>
> DCP-o-matic 1.78.0 was released. It is available from:
> http://dcpomatic.com/download
>
> This version includes:
>
> A new Russian translation from Igor Voytovich.
> An updated Italian translation from Davide Sanvito.
> An updated German translation from Carsten Kurz.
> An updated French translation from Thierry Journet.
> An updated Spanish translation from Manuel AC.
> Fix a regression causing “programming error” at the end of encodes in some cases.
> Fix one cause of hangs at the end of encodes (#508).
> Fix frame corruption and crashes when duplicating source frames in the DCP (e.g.
with 25fps content in a 50fps DCP).
> Fix incorrect timing of sound content when used with video that is being run
faster or slower than its default rate.
> Fix a couple of errors when trying to put 1920x1080 content into a Scope container
with no scale.
> Fix potential crash on encoding a second film in a session after closing the
first.
> Use the number of mapped audio channels for the DCP name (not the number of DCP
channels) (#479).
> Speculative fix for KDMs being rejected by EasyDCP player.
> Remove 1.37 ratio and correct Academy ratio to 1.375.
Use a less accurate but much less
time-consuming hashing method for content, dramatically speeding up examination of large
files.
> Fix snap to work for all edges (#473).
> Fix hiding of filters edit button when there are a lot of filters (#475).
> Allow the audio parts of combined video/audio files to be removed in the timeline,
which just unmaps the audio (#316).
> Only allow even DCP audio channel counts between 2 and 16 inclusive.
> Remove use of VRFY in SMTP sessions (#477).
> Examine a whole file to find its length if there is no readable header.
> Fix update checker.
>
> Thanks to Carsten Kurz, Igor Voytovich, Christian Hove, Jose Angel Velasco
> Fernandez, Steve Smail, Markus Raab, Davide Sanvito, Thierry Journet and
> Manuel AC.
>
> Six new supporters have been added since 1.77.0: thanks to Jose Angel
> Velasco Fernandez, Christopher Gray, Rui Pereira, Lawrence Towers, Frank
> Wenz and Roland Wirtz.
>
> Best regards,
> Carl_______________________________________________
> DCPomatic mailing list
> DCPomatic(a)carlh.net
> http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic
_______________________________________________
DCPomatic mailing list
DCPomatic(a)carlh.net
http://main.carlh.net/cgi-bin/mailman/listinfo/dcpomatic