DAG Report from Production Courses
The following data was captured on Feb 8, 2016 on a sandbox connected to Prod-Clone using the generate_course_blocks management command. The raw output from the management command is at DAG Report Raw Output.
Total Number of Courses that are DAGs: 63 out of 1751
Total Number of Blocks across all courses with multiple Parents: 1336
discussion: 720
html: 513
vertical: 53
customtag: 48
lti: 2
|
| # multi-parent blocks | discussion | html | vertical | customtag | other block types |
|---|---|---|---|---|---|---|---|
course-v1:MITx+6.00.1x_7+3T2015 | 2015 | 1 |
| 1 |
|
|
|
course-v1:MITx+6.004.2x+3T2015 | 2015 | 2 |
| 2 |
|
|
|
course-v1:MITx+8.MechCx_2+2T2015 | 2015 | 3 |
| 3 |
|
|
|
course-v1:MITx+6.004.1x_2+3T2015' | 2015 | 7 |
| 7 |
|
|
|
course-v1:Arbisoft+Lahore101+July2015 |
| 1 |
|
|
|
| lti |
BerkeleyX/CS184.1x/2013_Spring | 2013 | 10 |
| 10 |
|
|
|
course-v1:UCSDx+CSE167x+4T2015 | 2015 | 10 |
| 10 |
|
|
|
course-v1:MITx+6.002x_6x+1T2015 | 2015 | 94 | 62 | 31 |
| 1 |
|
course-v1:MITx+6.00.1x_6+2T2015 | 2015 | 1 |
| 1 |
|
|
|
MITx/7.00x_2/2T2014 | 2014 | 15 |
| 15 |
|
|
|
course-v1:MITx+6.002.Cx+3T2015 | 2015 | 25 | 12 | 13 |
|
|
|
MITx/6.00.1x/3T2013 | 2013 | 32 | 8 | 1 | 22 | 1 |
|
MITx/8.01x/2013_SOND | 2013 | 53 |
| 53 |
|
|
|
course-v1:MITx+7.00x_3+2T2015 | 2015 | 14 |
| 14 |
|
|
|
course-v1:MITx+8.05x_1+3T2015 | 2015 | 1 |
| 1 |
|
|
|
MITx/6.341x/3T2014 | 2014 | 1 |
|
|
|
| lti |
MITx/6.002x/2013_Spring | 2013 | 110 | 76 | 30 | 2 | 2 |
|
MITx/6.00.2x/1T2014 | 2014 | 6 | 4 | 1 |
| 1 |
|
MITx/8.EFTx/3T2014 | 2014 | 15 | 6 | 9 |
|
|
|
course-v1:UCSDx+CSE167x+3T2015 | 2015 | 10 |
| 10 |
|
|
|
aaa/bbb/ccc |
| 32 | 1 | 17 | 14 |
|
|
MITx/8.MReVx/2T2014 | 2014 | 2 | 1 | 1 |
|
|
|
MITx/2.01x/2T2014 | 2014 | 26 | 26 |
|
|
|
|
XuetangX/PerformanceTest101/2T2014 | 2014 | 110 | 77 | 31 |
| 2 |
|
BerkeleyX/CS169.2x/Software_as_a_Service |
| 5 | 4 |
| 1 |
|
|
MITx/7.QBWx/2T2014 | 2014 | 3 |
|
| 3 |
|
|
MITx/6.00.1-x/1T2014 | 2014 | 4 | 3 | 1 |
|
|
|
HarvardX/HLS1xD/Copyright |
| 1 | 1 |
|
|
|
|
course-v1:MITx+6.002x+2013 | 2013 | 110 | 76 | 30 | 2 | 2 |
|
MITx/EECS.6.002x/3T2013 | 2013 | 112 | 77 | 31 | 2 | 2 |
|
course-v1:MITx+6.00.1x_8+1T2016 | 2016 | 1 |
| 1 |
|
|
|
MITx/7.00x/2013_SOND | 2013 | 17 |
| 15 | 2 |
|
|
BerkeleyX/CS-184.1x/2013_October | 2013 | 10 |
| 10 |
|
|
|
edX/Test/3T2013 |
| 112 | 77 | 31 | 2 | 2 |
|
course-v1:MITx+6.00.2x_4+3T2015 | 2015 | 6 | 4 | 1 |
| 1 |
|
course-v1:LinuxFoundationX+LFS151x+2T2016 | 2016 | 2 |
| 2 |
|
|
|
MITx/6.832x/3T2014 | 2014 | 1 |
| 1 |
|
|
|
ak/akSState/1T2014 | 2014 | 8 |
| 8 |
|
Comments
Ugh. I thought I commented on this, but I guess I forgot to save it.
Can anyone help me understand the DAG blocks that aren't related to discussions, i.e the HTML block in course-v1:MITx+6.00.2x_3+1T2015 and the three verticals in MITx/7.QBWx/2T2014
How do those come about?
@(deprecated) Pinch (Unlicensed)In regards to "how" the multi-parented HTML blocks came about, since DAGs are not create-able in studio, they must have been authored via XML. It is a good question whether or not they are, in fact, intentional. I wasn't able to analyze "course-v1:MITx+6.00.2x_3+1T2015" since it has a "combinedopenended" block type, which we no longer support. However, in looking at another MIT course with multi-parented HTML blocks, it may have been created as an authoring error. For example, both units, Problem 4-1 and Problem 4-2 have the same HTML description: https://courses.edx.org/courses/MITx/6.00.1x_5/1T2015/courseware/Final/S2013_Final/
DAG alert - block-v1:MITx+6.00.2x_3+1T2015+type@html+block@exams:S2013_Final_Problem_4 has multiple parents: [u'block-v1:MITx+6.00.2x_3+1T2015+type@vertical+block@Problem_4-1', u'block-v1:MITx+6.00.2x_3+1T2015+type@vertical+block@Problem_4-2'].
Regarding verticals with multiple parents, I will rerun my updated script against prod-clone to see if I can get any additional information.
@Dave Ormsbee (Deactivated), @Piotr Mitros (Deactivated), @Peter Pinch, @Adam Palay (Deactivated), @Calen Pennington (Deactivated) Following up on the conversation from Why is the Course Structure a DAG, given the data captured above, I am not removing support for DAGs at the Block Transformers layer, as originally intended in this ticket https://openedx.atlassian.net/browse/MA-1339.
However, we continue to have development/performance costs in maintaining DAGs and the platform doesn't fully support it in all layers and all features.
DAGs are most heavily used by MIT, and that too, by older runs of courses. A few new courses also use DAGs - some probably because of legacy structures in reruns, while others for HTML blocks and discussions. For discussions, however, as @(deprecated) Pinch (Unlicensed)mentioned, they can now be updated since we have decouple discussion Topic IDs from block IDs.
I still feel we have to have a strong stance on this - either embrace them fully or remove them. If the former, we should work on improving their performance and fix all the areas of the code that use them. If the latter, we'll need to migrate new courses to use alternative ways of supporting their DAG requirements.
FYI @EdwardF (Deactivated).