nutch conda package is broken

[ahmadia]

The nutch conda package invokes a shell script that does some very funny Java hardcode things. This is fragile and we should get rid of it. Who is the maintainer of the nutch conda package?

[brittainhard]

Ben is. I created elasticnutch off of @quasiben's nutch package.

[ahmadia]

Another example. Error codes are not properly passed through these shell scripts.

[ahmadia]

@brittainhard - can you point me to the conda recipe for elasticnutch? We can raise an issue to move this over to a Salt install.

[brittainhard]

@ahmadia conda recipie is in the tar files. You can download them here: https://binstar.org/brittainhard/elasticnutch/files

[ahmadia]

Thanks, found it. @brittainhard - did you keep a record of how this is transformed over from a nutch install? Your recipe doesn't point to a specific commit from the Nutch repository.

[ahmadia]

(This is not high priority by the way, we can chat about it when I'm in ATX this week)

[brittainhard]

@ahmadia I didn't write that conda recipie, BTW, merely copied what was there. Not sure where you could figure out what version or commit that package pulls from. Couldn't have gotten anywhere fast if I had to. The best person to ask about that is Ben.

[ahmadia]

@brittainhard - there's a top-level recipe.json file that contains that sort of information. It's not mandatory to fill it out, so it can be hard to back out exactly where the files came from.

[chrismattmann]

This needs to be coordinated with @lewismc - he is going to try and maintain a conda package associated with Apache Nutch and the Nutch releases. Relevant issue: NUTCH-2025 I would love to see you guys working on this over there.

[quasiben]

Nutch recipe here: https://github.com/ContinuumIO/nutchpy/tree/master/conda-recipes/nutch_recipe

[lewismc]

Dynamite Ben that is fantastic.
Tomorrow I'm going to make sure I augment the release procedure to ensure
each subsequent Nuth release manager pushes the conda package as well!

On Tuesday, June 2, 2015, Benjamin Zaitlen notifications@github.com wrote:

Nutch recipe here:
https://github.com/ContinuumIO/nutchpy/tree/master/conda-recipes/nutch_recipe

—
Reply to this email directly or view it on GitHub
#531 (comment)
.

Lewis

[quasiben]

@ahmadia, if i recall correctly, the reason we built this funky shell script to call nutch and crawl is because the original binaries expect a specific path layout which didn't work nicely with conda envs. The reason out (though arguably short-sighted) was to leave nutch alone and just simply pass all commands from nutch in envs/.../bin to nutch in env/.../lib/nutch/bin . The other benefit we were thinking about at the time was localization of nutch and components in one place.