docs: add example for custom file format with COPY TO
#11174
Conversation
tshauck
changed the title
COPY TO
tshauck
changed the title
COPY TOCOPY TO
|
|
tshauck
force-pushed
the
add-example-for-copy-to
branch
from
ece7907 to
04d7fe7
Compare
tshauck
force-pushed
the
add-example-for-copy-to
branch
from
04d7fe7 to
c657653
Compare
tshauck
force-pushed
the
add-example-for-copy-to
branch
from
c657653 to
c1af955
Compare
| @@ -29,5 +29,6 @@ do | |||
| # Skip tests that rely on external storage and flight | |||
| if [ ! -d $filename ]; then | |||
| cargo run --example $example_name | |||
| cargo clean -p datafusion-examples | |||
There was a problem hiding this comment.
I ran into consistent build issues where the runner would run out of disk space. This is the best solution I could come up with. It seems to add about 5-10 seconds to the example action, which takes about 12 minutes overall.
Certainly open to alternatives.
There was a problem hiding this comment.
I am hitting similar issues with trying to add new examples in #11089
This seems like a good idea to me
My best hope is to move some of the example binaries into inline examples in the docs instead: #11178
Hopefully that will free up some additional space as well as make the examples easier to navigate
There was a problem hiding this comment.
er, sorry this came up. There are a few other tricks here (actions/runner-images#2840 (comment)) to clear up some disk space that may help also.
There was a problem hiding this comment.
No worries! I think this is a good excuse / reason to take another pass through the examples directory (and the library guide that you started however long ago).
There was a problem hiding this comment.
Thank you @tshauck for putting this example together! I think that wrapping an existing FileFormat was a clever way to demonstrate this without tons of boilerplate code being required for a working example. It ran locally for me no issues.
I left just one small suggestion.
There was a problem hiding this comment.
Thank you so much @tshauck -- I also think this looks really nice.
Thank you for the review @devinjdangelo
I agree the idea of wrapping an existing format to show the API is clever. I was thinking it would be awesome to somehow cook up a simple to implement custom format that would handle things entirely end to end but I can't think of any format suitable simple.
|
Yeah, I started down the road of something custom, but it turned out to be a non-trivial amount of code. Perhaps adding a link to something like https://github.com/datafusion-contrib/datafusion-orc would show a more complex example, though it doesn't implement |
|
Thanks again @tshauck and @devinjdangelo |
* feat: add example for copy to * better docs plus tempdir * build: clean examples if over 10GB * only 1GB * build: try clearing some disk space before running * build: remove sudo * build: try clean * build: run clean * build: only clean examples * docs: better output for example
Which issue does this PR close?
Closes #11079
Rationale for this change
Adds an example of for how to
COPY table TOa custom file format.What changes are included in this PR?
Created sort of a mock file format factory and file format that just wrap the CSV one but for a TSV.
Are these changes tested?
yes, I've run the example
Are there any user-facing changes?
no