Add NGSPlotDB-hg19-data recipe #9157

apeltzer · 2018-06-05T19:08:19Z

I have read the guidelines for bioconda recipes.
This PR adds a new recipe.
AFAIK, this recipe is directly relevant to the biological sciences (otherwise, please submit to the more general purpose conda-forge channel).
This PR updates an existing recipe.
This PR does something else (explain below).

The GDrive URL is somewhat a bit crazy - submitted a PR to get this to Galaxy Cargo-Port too. Maybe that eases things...

Referencing nf-core/chipseq#2
and nf-core/chipseq#15 to keep things related.

If this one is fine, I'm trying to do the same for hg38 and mm10 to cover the most used DBs for ngsplot.

apeltzer · 2018-06-06T08:28:05Z

@bioconda/core The r-ngsplot package seems to be missing for this? I submitted that yesterday, @bgruening merged it in - so it should be there right?

mbargull · 2018-06-06T08:36:34Z

You can remove the skip: True line from the r-ngsplot recipe (see https://github.com/bioconda/bioconda-recipes/pull/9149/files#r193333485) and add that change to this PR.

apeltzer · 2018-06-06T08:46:03Z

Whoopsies!

apeltzer · 2018-06-06T08:51:15Z

Hmm, doesn't build that way. I will probably resubmit in a different PR then?! @mbargull

mbargull · 2018-06-06T08:40:11Z

recipes/r-ngsplot-hg19/meta.yaml

+  sha256: a3ad6daceec383f88faf3d3ee899f2bef37b3be2658ee9afa01e86404c0c92bd
+build:
+  number: 0
+  rpaths:


You can remove rpaths for a data package.

mbargull · 2018-06-06T08:40:37Z

recipes/r-ngsplot-hg19/meta.yaml

+    - lib/R/lib/
+    - lib/
+requirements:
+  build:


No build requirement needed for a data package.

Thanks, removed that

mbargull · 2018-06-06T08:50:13Z

recipes/r-ngsplot-hg19/post-link.sh

+
+  # Platform-specific md5sum checks.
+  if [[ $(uname -s) == "Linux" ]]; then
+    if md5sum -c <<<"$MD5  $TARBALL"; then


If you use a sha256 hash, you'll have to use sha256sum here.
But note that on macOS you'll likely have to use another command for this. Can't remember what, though.
The safe route might be to add a dependency on openssl and use openssl sha256? (E.g.,

bioconda-recipes/recipes/bamutil/build.sh

Line 5 in f5c3727

openssl sha256 libStatGen-v$VERSION.tar.gz | grep 70a504c5cc4838c6ac96cdd010644454615cc907df4e3794c999baf958fa734b

)

Changed back to md5

mbargull · 2018-06-06T08:50:29Z

recipes/r-ngsplot-hg19/post-link.sh

+      break
+    fi
+  else if [[ $(uname -s) == "Darwin" ]]; then
+    if [[ $(md5 $TARBALL | cut -f4 -d " ") == "$MD5" ]]; then


sha256, see above

Changed to md5

mbargull · 2018-06-06T08:53:15Z

recipes/r-ngsplot-hg19/meta.yaml

+package:
+  name: 'ngsplotdb-{{ name|lower }}'
+  version: '{{ version }}'
+source:


No source needed since post-link.sh does the job.

Ok, removed that - but the md5 is still required / good practice right?

Nah, the md5 in meta.yaml is only tied to the one in source, hence can be removed, too.

mbargull · 2018-06-06T09:08:22Z

recipes/r-ngsplot-hg19/post-link.sh

+fi
+
+# Install and clean up
+ngsplotdb.py install ngsplotdb_hg19_75_3.00.tar.gz


You want to use "${TARBALL}" here.
Also, if this outputs anything (progress status and such), you'll want to write that to .messages.txt, see https://conda.io/docs/user-guide/tasks/build-packages/link-scripts.html, i.e.,

ngsplotdb.py install "${TARBALL}" > "${PREFIX}/.messages.txt"

maybe even

ngsplotdb.py install "${TARBALL}" > "${PREFIX}/.messages.txt" 2>&1

to also capture stderr.

mbargull · 2018-06-06T09:08:46Z

recipes/r-ngsplot-hg19/pre-unlink.sh

@@ -0,0 +1 @@
+ngsplotdb.py remove hg19


Redirect output to .messages.txt, see above.

mbargull · 2018-06-06T09:13:33Z

recipes/r-ngsplot-hg19/post-link.sh

+
+if [[ $SUCCESS != 1 ]]; then
+  echo "ERROR: post-link.sh was unable to download any of the following URLs with the md5sum $MD5:"
+  printf '%s\n' "${URLS[@]}"


I think even this should write to .messages.txt (see below), i.e.,

echo "ERROR: post-link.sh was unable to download any of the following URLs with the md5sum $MD5:" > "${PREFIX}/.messages.txt" 2>&1 printf '%s\n' "${URLS[@]}" > "${PREFIX}/.messages.txt" 2>&1

mbargull · 2018-06-06T09:21:43Z

Hmm, doesn't build that way. I will probably resubmit in a different PR then?! @mbargull

You can use this one, no problem.

+#!/bin/bash
+outdir=$PREFIX/share/$PKG_NAME-$PKG_VERSION-$PKG_BUILDNUM
+mkdir -p $outdir
+mkdir -p $PREFIX/bin
+cp -R * $outdir

^-- please don't copy everything unconditionally but rather selectively the files you want.

+#Set up links for 
+for f in $outdir/*; do
+    ln -s $outdir/* $PREFIX/bin

^-- you'll want to use ln -s "${f}" "${PREFIX}/bin" here. But also, you should not loop over $outdir/* but just the files in $outdir that are the executables you want to link to.

+    fbname=$(basename "$f")
+    chmod 0755 ${PREFIX}/bin/$fbname
+done

mbargull · 2018-06-06T09:29:14Z

I took a quick peek at the file list in https://github.com/shenlab-sinai/ngsplot/archive/2.63.tar.gz. There is a bin folder in there => I assume you want to only link the files (or possibly only a subset of them?) in "${outdir}"/bin/* then, right?

mbargull · 2018-06-06T09:31:11Z

And the files in bin seem to have the executable mode set already, to the chmod is likely not needed.

apeltzer · 2018-06-06T10:56:44Z

Hi @mbargull !
thanks for all the tips / hints. The problem is, that the respective tools I need in ngsplot.py use system calls to other tools in the bin folder and I don't want to try & error each of them until the recipe works. Honestly, I also feel like this package is rather not too well prepared in terms of keeping things easy to install for end users, which is also the main reason for packaging this nicely (or as nice as it gets at least).

Therefore, I'd rather say I restrict to simply add all tools in bin to the path, similar to what the authors suggest in their readme . What do you think?

mbargull

A few changes needed, but it looks nearly done 🎉.

r-ngsplot built fine, but I'm not sure whether the tests actually succeeded. By piping to true, you're masking the failures, I think (that is, I don't believe conda-build runs the tests with -o pipefail). Please remove all | true.
If the programs return non-zero exit codes on success, a workaround is to pipe to grep -q 'some expression you know the program outputs.

Therefore, I'd rather say I restrict to simply add all tools in bin to the path, similar to what the authors suggest in their readme . What do you think?

Seems fine to me.

mbargull · 2018-06-06T12:19:01Z

recipes/r-ngsplot-hg19/meta.yaml

+    - wget
+test:
+  commands:
+    - ngsplotdb.py install "${TARBALL}" > "${PREFIX}/.messages.txt" 2>&1


The test is run after the package is installed, i.e., after the post-link.sh is run.
So, that ngsplotdb.py install is already done at that point. Rather than executing it again, you should run a test that checks if the database has been successfully installed (either with some (short-running!) command from r-gnsplot or if that's not available, at least something like test -f "${PREFIX}/some-file-path".)

mbargull · 2018-06-06T12:19:13Z

recipes/r-ngsplot-hg19/meta.yaml

+package:
+  name: 'ngsplotdb-{{ name|lower }}'
+  version: '{{ version }}'
+  md5: 172a6a93de3d1ae8a3ad6d6d79e911f7


You can remove that md5 here.

mbargull · 2018-06-06T12:20:12Z

recipes/r-ngsplot-hg19/post-link.sh

+URLS=(
+  "https://drive.google.com/uc?export=download&id=0B5hDZ2BucCI6SURYWW5XdUxnbW8"
+)
+sha256="a3ad6daceec383f88faf3d3ee899f2bef37b3be2658ee9afa01e86404c0c92bd"


MD5='172a6a93de3d1ae8a3ad6d6d79e911f7'

instead of sha256

mbargull · 2018-06-06T12:22:30Z

recipes/r-ngsplot-hg19/post-link.sh

+
+SUCCESS=0
+for URL in ${URLS[@]}; do
+  wget -O- -q $URL > $TARBALL


The CI choked on the unquoted $URL, I think.

wget -qO "${TARBALL}" "${URL}"

should do the trick.

mbargull · 2018-06-06T12:23:48Z

recipes/r-ngsplot-hg19/post-link.sh

+done
+
+if [[ $SUCCESS != 1 ]]; then
+  echo "ERROR: post-link.sh was unable to download any of the following URLs with the md5sum $MD5:" > "${PREFIX}/.messages.txt" 2>&1


My bad, should be (appending to .messages.txt):

echo "ERROR: post-link.sh was unable to download any of the following URLs with the md5sum $MD5:" >> "${PREFIX}/.messages.txt" 2>&1

mbargull · 2018-06-06T12:25:37Z

recipes/r-ngsplot-hg19/pre-unlink.sh

@@ -0,0 +1 @@
+ngsplotdb.py remove hg19 > "${PREFIX}/.messages.txt" 2>&1


Also, missing new line on EOF 😉

mbargull · 2018-06-06T12:25:44Z

recipes/r-ngsplot-hg19/post-link.sh

+# Install and clean up
+ngsplotdb.py install "${TARBALL}" > "${PREFIX}/.messages.txt" 2>&1
+rm $TARBALL
+rmdir $STAGING


Missing new line on EOF 😉

mbargull · 2018-06-06T12:26:24Z

recipes/r-ngsplot/build.sh

-for f in $outdir/*; do
-    ln -s $outdir/* $PREFIX/bin
+for f in ${outdir}/bin/*; do
+    ln -s ${f} ${PREFIX}/bin
    fbname=$(basename "$f")


Can be removed now.

I don't get it right now - why?

Sorry, I've been imprecise. Just meant the fbname=$(basename "$f") part 😉

mbargull · 2018-06-06T12:26:31Z

recipes/r-ngsplot/build.sh

    fbname=$(basename "$f")
-    chmod 0755 ${PREFIX}/bin/$fbname
 done


Also, missing new line on EOF 😉

mbargull · 2018-06-06T12:31:08Z

recipes/r-ngsplot/build.sh

@@ -4,8 +4,7 @@ mkdir -p $outdir
 mkdir -p $PREFIX/bin
 cp -R * $outdir 


I'm not sure if it's still the case, but I know conda-build sometimes wrote files to the cwd during building (I think conda-build.sh or run-test.sh or something). Those files will/might get copied too if you cp -R * here.

mbargull · 2018-06-06T12:41:48Z

recipes/r-ngsplot-hg19/post-link.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 FN="ngsplotdb_hg19_75_3.00.tar.gz"
 URLS=(
-  "https://doc-0g-a0-docs.googleusercontent.com/docs/securesc/ha0ro937gcuc7l7deffksulhg5h7mbp1/jiksn2m41cm3dbspq0nqt102h6m6eo1o/1528221600000/01382619737792242945/*/0B5hDZ2BucCI6SURYWW5XdUxnbW8?e=download"
+  "https://drive.google.com/uc?export=download&id=0B5hDZ2BucCI6SURYWW5XdUxnbW8"


👍 looks much saner 😄. galaxyproject/cargo-port#148 is probably still preferred, I think, but if that isn't merged in a timely manner, we can always package a new build for r-ngsplot-hg19 updating that URL later.

apeltzer · 2018-06-08T11:56:32Z

So setting this environment variable as suggested in the README won't work. Therefore, the second test fails and probably also the recipe can't be called properly.

Using something like this:

#Setting environment variable NGSPLOT as requested by authors readme
ln -s ${outdir} ${PREFIX}/bin/NGSPLOT

Doesn't work unfortunately... Setting something via export NGSPLOT=/path/to... doesn't work either.

@mbargull any idea? This is the last thing required for the recipe...

apeltzer · 2018-06-08T12:22:51Z

Thanks @ewels for your hint with https://github.com/bioconda/bioconda-recipes/blob/master/recipes/bcftools/1.2/build.sh, this seems to work locally now at least!

apeltzer · 2018-06-08T12:42:34Z

Ok, I don't have a clue what causes this test fail...

@bioconda/core anybody an idea?

mbargull · 2018-06-08T13:02:20Z

recipes/r-ngsplot/build.sh

-for f in $outdir/*; do
-    ln -s $outdir/* $PREFIX/bin
+for f in ${outdir}/bin/*; do
+    ln -s ${f} ${PREFIX}/bin
    fbname=$(basename "$f")


Sorry, I've been imprecise. Just meant the fbname=$(basename "$f") part 😉

mbargull · 2018-06-08T13:06:10Z

recipes/r-ngsplot/build.sh

+mkdir -p $DEACTIVATE_DIR
+cp $RECIPE_DIR/scripts/activate.sh $ACTIVATE_DIR/ngsplot-activate.sh
+cp $RECIPE_DIR/scripts/deactivate.sh $DEACTIVATE_DIR/ngsplot-deactivate.sh


To avoid conflicts, you should rename those to the exact name of the package, i.e., r-ngsplot-.... You can use the ${PKG_NAME} variable for this, see, e.g.: https://github.com/conda-forge/toolchain-feedstock/blob/bf0bfa1f06f6592eb3e6ef8989339a005cfd93ef/recipe/install-toolchain.sh#L8

mbargull · 2018-06-08T13:07:31Z

recipes/r-ngsplot-hg19/post-link.sh

+
+if [[ $SUCCESS != 1 ]]; then
+  echo "ERROR: post-link.sh was unable to download any of the following URLs with the md5sum $MD5:" >> "${PREFIX}/.messages.txt" 2>&1
+  printf '%s\n' "${URLS[@]}" > "${PREFIX}/.messages.txt" 2>&1


Please fix my > mistake 😉

printf '%s\n' "${URLS[@]}" >> "${PREFIX}/.messages.txt" 2>&1

mbargull · 2018-06-08T13:07:42Z

recipes/r-ngsplot-hg19/post-link.sh

+fi
+
+# Install and clean up
+ngsplotdb.py install "${TARBALL}" > "${PREFIX}/.messages.txt" 2>&1


Please fix my > mistake 😉

ngsplotdb.py install "${TARBALL}" >> "${PREFIX}/.messages.txt" 2>&1

mbargull · 2018-06-08T13:08:05Z

recipes/r-ngsplot-hg19/pre-unlink.sh

@@ -0,0 +1 @@
+ngsplotdb.py remove hg19 > "${PREFIX}/.messages.txt" 2>&1


Please fix my > mistake 😉

ngsplotdb.py remove hg19 >> "${PREFIX}/.messages.txt" 2>&1

mbargull · 2018-06-08T13:29:37Z

Ok, I don't have a clue what causes this test fail...

That's because of a shortcoming on how the containers are built/run, see galaxyproject/galaxy-lib#81.
Essentially, the activate scripts are not run for the containers.

Choices you have for this:

"don't care" approach: Selectively disable the failing tests for the container by extracting them into run_tests.sh
"hero" approach: Fix activate.sh scripts should be converted into .bashrc during container build galaxyproject/galaxy-lib#81 😉
"patch" approach: If NGS_PLOT always points to that path, patch the scripts to use that path instead (probably not by using source/patches in meta.yaml but patching them in build.sh to be able to use the build-time environment variables the path consists of)
"defensive patch" approach: Just like "patch approach" but not replace NGS_PLOT with the path directly but replace it by ${NGS_PLOT:-SOME_PATH} (for shell scripts -- no idea what the R equivalent is).

mbargull · 2018-06-08T13:33:28Z

recipes/r-ngsplot/scripts/activate.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+export NGSPLOT=$PREFIX/share/$PKG_NAME-$PKG_VERSION-$PKG_BUILDNUM


Oh, and using $PKG_NAME-$PKG_VERSION-$PKG_BUILDNUM should not work here, I believe.
Those variable are only defined during the build/test phase or during post-link (and others), but not when activating an environment. If the tests passed for this, that's only because they were set in that case.
You'll want to create/modify the activation script in build.sh.

mbargull · 2018-06-08T13:38:48Z

recipes/r-ngsplot/build.sh

+mkdir -p $ACTIVATE_DIR
+mkdir -p $DEACTIVATE_DIR
+cp $RECIPE_DIR/scripts/activate.sh $ACTIVATE_DIR/ngsplot-activate.sh


In regards to the comment below for scripts/activate.sh:
Replace

cp $RECIPE_DIR/scripts/activate.sh $ACTIVATE_DIR/ngsplot-activate.sh

by something like

echo > "${ACTIVATE_DIR}/${PKG_NAME}-activate.sh" <<EOF #!/bin/bash export NGSPLOT="${outdir}" export _CONDA_SET_NGSPLOT_ENV=1 EOF

. ${outdir} will then have been evaluated at build time (becaus of using <<EOF with EOF being unquoted) and replaced by the correct output path (which conda will in turn adjust to the environment's path at install time).

mbargull · 2018-06-08T14:02:04Z

Ah, I feel the urge to clarify the

Choices you have for this

The different approaches are of course not necessarily exclusive -- "hero" is always something you can try to achieve 😄.

The "don't care" approach is not really preferable since it leaves the container in an incomplete/not directly usable state: The user would have to "guess" the path and set the variable themselves to make use of the container.

The "defensive patch" approach might be overly wary, but it retain's the software's functionality to let the user choose/set an alternative path. Probably that variable is just used as a workaround to achieve relocatability so that functionality isn't needed if you patch and let conda do this job -- BUT I don't know the program/scripts/uses at all and might be wrong about this. (As a general note, patching should always be done carefully, especially if it's changing the sofware's behavior. But it could be the best solution in this case.)

apeltzer · 2018-06-08T14:12:23Z

Thanks for all these hints - never had that much trouble with a single recipe than this one (which is tbh in a bad shape for packaging...).

I'll patch the call in the script I'm calling here in build.sh and the tool should then be able to function.

progpath <- Sys.getenv('NGSPLOT')
if(progpath == "") {

As far as I see it, this is the best way other than the hero approach, which I don't feel qualified to do.

apeltzer · 2018-06-08T14:27:40Z

So replaced the NGSPLOT ENV call in the R script. Lets hope that works fine then :-)

mbargull · 2018-06-09T06:50:03Z

Did a quick grep for NGSPLOT and it seems the code pattern for that is quite uniform. If the commit I added work, we can leave out the activation script entirely. Let's see how it goes

mbargull · 2018-06-09T17:01:10Z

Let's see how it goes

Didn't quite work yet... debugging...

Thanks for all these hints - never had that much trouble with a single recipe than this one (which is tbh in a bad shape for packaging...).

(I must've missed that comment before.) You're welcome, happy to help. Yeah it's a bit troublesome that it's not packaged as usual, but hey, you got to learn a few new tricks which is never a bad things 😉 .

mbargull · 2018-06-09T18:09:04Z

I think the last changes might make this pass. If it indeed does and you still think the patching is fine, feel free to merge!

mbargull · 2018-06-09T18:10:25Z

recipes/r-ngsplot-hg19/meta.yaml

+    - (ngsplotdb.py list | grep -qF "hg19" || ngsplotdb.py list)
+about:
+  home: 'https://github.com/shenlab-sinai/ngsplot'
+  license: GPL-2.0


I wonder: does the database have another home URL? Is the database itself also licensed under the GPL-2.0?

No, they are on GDrive and linked from the same Github repository - so I think its fine to use the same licence (as they didn't state explicitly another one).

Thanks for all the help - learned a lot here and will add two more databases today to finalize things here :-)

Alexander Peltzer added 2 commits June 5, 2018 21:05

Add NGSPlotDB-hg19-data recipe

f0180d7

Okay, lets test this

f4d66f2

Add fix for ngsplot recipe

1975340

mbargull requested changes Jun 6, 2018

View reviewed changes

Alexander Peltzer added 4 commits June 6, 2018 12:27

Removed

e9bd34b

Added @mbargull 's suggestions

3d214fb

More suggestions

e486e49

Add suggestions of @mbargull

f5af8a9

Alexander Peltzer added 4 commits June 6, 2018 12:58

Adjusted build.sh script

3ead683

Removed source, as this is provided by post-link

b1ae475

Hopefully more permant file URL

cf50794

Double check

ae6e95a

mbargull reviewed Jun 6, 2018

View reviewed changes

Add check for hg19 database, removed sha256

1590814

mbargull reviewed Jun 6, 2018

View reviewed changes

Alexander Peltzer added 7 commits June 6, 2018 21:38

Better tests for ngsplot

e4eb39e

More specific copying

721e3a2

Post and Pre Unlink scripts for r-ngsplot-hg19 updated

054a918

No need for images in there

a76b668

Fixing tests

2a7f5e4

Fixing tests

484b1f2

Updated Build Script

e16b597

Alexander Peltzer added 2 commits June 8, 2018 14:17

Add Activate / Deactivate scripts for ENV variable

95efbc3

Wrong path

9c44497

mbargull requested changes Jun 8, 2018

View reviewed changes

mbargull reviewed Jun 8, 2018

View reviewed changes

Removing unrequired fbname

b35f91a

Alexander Peltzer added 2 commits June 8, 2018 16:04

Add exact name of package

05101fa

Fixing pipes

8fb135d

Alexander Peltzer added 2 commits June 8, 2018 16:26

Get rid of old activate script

30d2a97

Add patch for environment variable in R script

9b168e6

mbargull force-pushed the r-ngsplot-hg19 branch from 3273bf3 to 7977757 Compare June 9, 2018 06:48

mbargull added 2 commits June 9, 2018 08:50

r-ngsplot: patch all scripts, avoid activation script

7977757

r-ngsplot: debugging: show output if test failed

a40cf32

mbargull added 2 commits June 9, 2018 19:29

r-ngsplot: try || in subshells

f097068

r-ngsplot: add missing python dependency

de5d1a6

mbargull approved these changes Jun 9, 2018

View reviewed changes

r-ngsplot-hg19: fix variable name in post-link.sh

88af4ed

mbargull reviewed Jun 9, 2018

View reviewed changes

apeltzer merged commit f36bfb7 into master Jun 11, 2018

mbargull deleted the r-ngsplot-hg19 branch June 11, 2018 07:07

		@@ -0,0 +1 @@
		ngsplotdb.py remove hg19 > "${PREFIX}/.messages.txt" 2>&1

		@@ -4,8 +4,7 @@ mkdir -p $outdir
		mkdir -p $PREFIX/bin
		cp -R * $outdir

		@@ -0,0 +1,3 @@
		#!/bin/bash
		export NGSPLOT=$PREFIX/share/$PKG_NAME-$PKG_VERSION-$PKG_BUILDNUM

Add NGSPlotDB-hg19-data recipe #9157

Add NGSPlotDB-hg19-data recipe #9157

Conversation

apeltzer commented Jun 5, 2018

apeltzer commented Jun 6, 2018

mbargull commented Jun 6, 2018 • edited Loading

apeltzer commented Jun 6, 2018

apeltzer commented Jun 6, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mbargull commented Jun 6, 2018

mbargull commented Jun 6, 2018

mbargull commented Jun 6, 2018

apeltzer commented Jun 6, 2018 • edited Loading

mbargull left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

apeltzer commented Jun 8, 2018

apeltzer commented Jun 8, 2018

apeltzer commented Jun 8, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mbargull commented Jun 8, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mbargull commented Jun 8, 2018

apeltzer commented Jun 8, 2018

apeltzer commented Jun 8, 2018

mbargull commented Jun 9, 2018

mbargull commented Jun 9, 2018

mbargull commented Jun 9, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mbargull commented Jun 6, 2018 •

edited

Loading

apeltzer commented Jun 6, 2018 •

edited

Loading