SGF download bugs

I noticed a few bugs and inconsistencies with an SGF file downloaded from a review. Some of these issues appear to be present with all SGF files downloaded from OGS in general.

This is the review in question:

Here are the issues:

  1. The downloaded SGF file has several more variations that are not visible on the website. For example, see the variations for moves 195, 191, 57, 55, 54, 32 (and maybe others as well) that are present in the SGF file, but are missing when viewing the review online.
  2. The comment in the first node (move 0) is missing in the downloaded SGF file.
  3. The generated file duplicates properties rather using a single property with a list of values, which violates the SGF standard and may cause issues when opening with other software. For example, OGS uses “TR[me]TR[ne]” or “LB[cn:A]LB[dp:C]LB[cr:B]” where it should use “TR[me][ne]” or “LB[cn:A][dp:C][cr:B]” instead.
  4. In the plaintext of the SGF file there is a newline at the end of every comment (before the closing square bracket). This is just a minor formatting issue and not even technically against the SGF standard, but it seems unnecessary and creates a strangely formatted file when there are also two newlines after the closing square bracket of every comment.
1 Like

Just found this other related post:

1 Like

Thanks! Added to this list o stuff to fix


I also noticed that all of the SGF files (for both game records and review) seem to have the copyright property set to “CP[]”, which would imply the OGS holds copyright on all of the game records and reviews.

Is this really appropriate? I’m not sure if copyright is even applicable to the plain game records (minus annotations and comments), and the reviews would seem to be the creative work of the reviewer (and hence, in many jurisdictions, subject to automatic copyright belonging to the reviewer).

The OGS terms of service ( even states that “ does not claim ownership of any Content you submit or otherwise make available through the Service”, and instead just mentions an implied license to allow for redistribution.

Perhaps using the source property (e.g., SO[]) would be more appropriate.

Also, for reviews, it would be nice to note the reviewer using the annotator property (e.g., AN[reviewer name]).


I’d vote for this suggestion.

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.