reproducibility - Githubissues

wcornwell commented 6 years ago

@willpearse a couple OSX issues:

right now on OSX the clean_data/gbif_tpl_locations.txt is coming out as an empty file. Not totally sure why.
clean_data/gbif_spp.txt still has authorities, unlike clean_data/genbank_spp_clean.txt and clean_data/try_spp_clean.txt

wcornwell commented 6 years ago

Reproducibility is hard!

on centOS (ruby/2.3.7) single quotes in the rake file don't work, but double quotes do. argggg

willpearse commented 6 years ago

No idea about why gbif_tpl_locations.txt is coming out empty for Mac. I'll try and borrow Bonnie's laptop this evening and have a look, but I'm not sure what's going on there.
I'm not so surprised about the authority stuff; gbif_spp.txt is purely a list of unique names (with authorities), which is used in a few other scripts (and I thought could be useful for the Venn). That's a "feature" not a "bug"
That's infuriating about the single quotes, but at least that's an easy fix!...

willpearse commented 6 years ago

I think I might have found the problem: that's not a single quote, it's a backtick. It's Ruby-speak for "run the following shell command". The change you've just made should stop genbank_raw_names from every being correctly produced. If you're talking about the changes in the R script section (later in the file, not where I've highlighted), then ignore me.

I should have thought of this earlier. This is one of the most useful things in Ruby, but it's also wildly non-obvious unless you're told about it. Sorry!...

wcornwell commented 6 years ago

OK. my bad on the backticks...learning....

cool about the authorities in gbif_spp.txt just need to strip them out somewhere--at some point--to get the overlap calculations to work again. Can be later in the workflow though if the authorities are useful for some things.

what happens between gbif_cut.txt and gbif_tpl_locations.txt? Something there isn't working for me, but without returning an error. It does take a while to think about it though and ruby is running at 100% on one core, so I feel like it's something about the writing-to-file part.

wcornwell commented 6 years ago

gbif_tpl_locations.txt also 0 B on linux, so not an operating system thing. Maybe a ruby version thing?

willpearse commented 6 years ago

No worries about the back-ticks. Again, it's really more my fault for not explaining what was going on, and shifting to Ruby in the first place.

I'm having similar problems with this bit of code. gbif_tpl_locations.txt is, I hope, going to be the file with synonymy handled. It's working for the first 100 or so species, but not for everything. I'm getting the same error; I pushed the code I was trying, if that makes sense?

Please leave this with me; I'm dropping in and out of this over the week as I set things running and then see what happens. I'm not sure what's going wrong as I scale up the number of species but I will figure it out...

traitecoevo / sTEP_overlap

reproducibility #9