Hi Lance,

I need to figure out precisely what steps you are taking to produce this behavior, as I have not been able to do so.  Please see my inline comments, and let me know more information about each step you are taking to produce this behavior.


On Sep 25, 2012, at 10:24 AM, Lance Parsons wrote:

I'm sorry I wasn't more clear.  I do believe that those links explain the behavior I am seeing. However, let me try to describe it a different way.  It seems that there will, at most, be one installable revision of a given version of a tool.  Here I use revision to denote a mercurial revision and version to denote the version described in the tool's config xml.  However, while it seems this is the desired behavior, it seems to lead to an undesirable situation in some circumstances.  Consider this scenario:

1) Upload version 0.1 of MYTOOL at revision 0:XXXXXXXX to the tool shed.

I assume you are uploading the initial tarball or separate files from a certain local directory.  Let's call it   TOOL_DEV_LOCAL_DIR.

2) Install version 0.1 of MYTOOL at revision 0:XXXXXXX to my local Galaxy instance.

This now creates a new mercurial repository at a specified location on your Galaxy server, let's call it  REPO_INSTALL_DIR.


3) Modify MYTOOL, leaving the version at 0.1,


You are doing the above step in TOOL_DEV_LOCAL_DIR, correct?


but changing the revision to 1:YYYYYYYY.


The tool shed is creating the new change set hash for you when you push the changes from TOOL_DEV_LOCAL_DIR to the repository in the tool shed, correct?


These updates would be for things that don't change the input or output of a tool (such as updates to documentation, the addition of tool dependencies, etc.)


OK, this make sense.


4) Attempt to update MYTOOL from my local Galaxy instance. 


You  are using the Galaxy UI to "Get updates" to your cloned repository, which is located in REPO_INSTALL_DIR, correct?


This now results in an exception: " RepoLookupError: unknown revision '1' ".


If answers to all of my above questions are "Yes", then this is where I cannot reproduce the behavior you are seeing.  If you are actually adding changsets to your cloned repository located in REPO_INSTALL_DIR, then pushing them to the repository in the tool shed, then attempting to get updates from the tool shed to your cloned repository in REPO_INSTALL_DIR, then this process may be causing the exception you are seeing.  You should not be adding changesets to the cloned repository, but to your local development environment in TOOL_DEV_LOCAL_DIR.


Deactivating and/or uninstalling MYTOOL works, but any attempt to reinstall the latest version results in an error stating that an older revision of the tool was installed.  I now have to way to (easily) update my local Galaxy instance to this updated revision of the tool.

My only workaround to this was to increment the version of MYTOOL and push another revision. This allowed me to install the new version and deactivate the old version in my local Galaxy instance.


As for the mercurial issue, I'm running OS X, using the Homebrew version of Mercurial:
hg --version
Mercurial Distributed SCM (version 2.3.1)
(see http://mercurial.selenic.com for more information)

Copyright (C) 2005-2012 Matt Mackall and others
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.


I'm running version 2.2.3, so I'll upgrade to version 2.3.1 to see if the newer version breaks my environment.



Let me know if I can help track things down any further, etc.  Thanks.

Lance

Greg Von Kuster wrote:
Hi Lance,


On Sep 21, 2012, at 6:04 PM, Lance Parsons wrote:

OK, I was able to get a new version installed.  It seems there are two issues:

1) New revisions with the same version "ionvalidate" previous revisions.  This means that Galaxy servers with the old, and now invalid, revisions are not able to update the tool (nor install it again).

I'm not quite sure what you're stating here.  Do the following tool shed wiki page clarify the behavior you are seeing?




2) Pushes from Mercurial (even version 2.3.3) do not seem to trigger metadata refreshes in the tool shed, however, uploads of tar.gz files do.

I am not able to reproduce this behavior.  In my environment, metadata is always automatically generated for new changesets I push to my local tool shed (or the test tool shed) from the command line.

What is the result of typing the following in the environment from which you are pushing changes to the tool shed?

$hg --version

You should see something like the following, showing that you are running at least hg version 2.2.3.

gvk:/tmp/repos/convert_chars gvk$ hg --version
Mercurial Distributed SCM (version 2.2.3)
(see http://mercurial.selenic.com for more information)

Copyright (C) 2005-2012 Matt Mackall and others
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.



Hope this helps. 

Lance

Lance Parsons wrote:
I've run into this issue again, and I'm having a hard time working around it.  However, I have confirmed that at least some updates to a tool in the tool shed will "invalidate" previously valid revisions and thus prevent users from installing or updating the tool at all.

For example, push version 0.1 of the tool and create a valid revision 1:xxxxxx.  Then install the tool in galaxy.  Make a small change (say to tool_dependencies.xml) and push a new revision (but keep the tool version the same), now at revision 2:xxxxxxx.  The tool shed will show 2:xxxxxx as the only valid revision to install, but the galaxy system with revision 1:xxxxxx will be stuck, unable to get upgrades (Server Error described previously). 

I'm trying to work around this now with my htseq-count tool, but so far no luck.  I've created a few spurious revisions in the attempt, and I think now I may just try bumping the version (already did to no avail, toolshed still thinks it's the same) and uploading a tar file.  That seems to more reliably parse metadata.  Will let you know what, if anything, works.  Thanks.

Lance

Greg Von Kuster wrote:
Hello Lance,

I've just committed a fix for getting updates to installed tool shed repositories in change set 7713:23107188eab8, which is currently available only in the Galaxy central repository.  However, my fix will probably not correct the issue you're describing, and I'm still not able to reproduce this behavior.  See my inline comments...


On Sep 13, 2012, at 4:41 PM, Lance Parsons wrote:

Actually, I think that is exactly the issue.  I DO have 3:f7a5b54a8d4f installed. I've run into a related issue before, but didn't fully understand it.

I believe what happened was:
1) I pushed revision 3:f7a5b54a8d4f to the tool shed which contained the first revision of version 0.2 of the htseq-count tool.
2) I installed the htseq-count tool from the tool shed, getting revision 3:f7a5b54a8d4f
3) I pushed an update to version 0.2 of the htseq-count tool. The only changes were to tool-dependencies so I thought it would be safe to leave the version number alone (perhaps this is problem?)


You are correct in stating that the tool version number should not change just because you've added a tool_dependencies.xml file.  This is definitely not causing the behavior you're describing.


4) I attempted to get updates and ran into the issue I described.

I also ran into this (I believe it was with freebayes, but not sure) when I removed (uninstalled) a particular revision of a tool. Then the tool was updated. I went to install and and it said that I already had a previous revision installed and should install that. However, I couldn't since the tool shed won't allow installation of old revisions of the same version of a tool.

The following section of the tool shed wiki should provide the details about why you are seeing this behavior.  Keep in mind that you will only get certain updates to installed repositories from the tool shed.  This behavior enables updates to installed tool versions.  To get a completely new version of an installed tool (if one exists), you need to install a new (different) changeset revision from the tool shed repository.




Let me know if there is anything I can do to help sort this out.

Lance

Greg Von Kuster wrote:
Hi Lance,

What is the changeset revision that you installed?  It looks like you could only have installed one of the following 3 revisions:


Since you could not have installed 3:f7a5b54a8d4f, I'm not quite sure how you could be trying to update to 4.  Did you install 4 and are trying to get updates?  

I've tried several things but am not able to reproduce this behavior, so it's difficult to determine what may be causing the problem

Greg Von Kuster

On Sep 12, 2012, at 3:08 PM, Lance Parsons wrote:

I've updated my development system now, and when I try to get updates for that particular tool (htseq_count) I run into the following error.  Any ideas on how I can/should fix this?  Thanks.

URL: http://galaxy-dev.princeton.edu/admin_toolshed/update_to_changeset_revision?tool_shed_url=http://toolshed.g2.bx.psu.edu/&name=htseq_count&owner=lparsons&changeset_revision=f7a5b54a8d4f&latest_changeset_revision=14bec14f4290&latest_ctx_rev=4
File '/data/galaxy-dev/galaxy-dev/eggs/WebError-0.8a-py2.6.egg/weberror/evalexception/middleware.py', line 364 in respond
app_iter = self.application(environ, detect_start_response)
File '/data/galaxy-dev/galaxy-dev/eggs/Paste-1.6-py2.6.egg/paste/debug/prints.py', line 98 in __call__
environ, self.app)
File '/data/galaxy-dev/galaxy-dev/eggs/Paste-1.6-py2.6.egg/paste/wsgilib.py', line 539 in intercept_output
app_iter = application(environ, replacement_start_response)
File '/data/galaxy-dev/galaxy-dev/eggs/Paste-1.6-py2.6.egg/paste/recursive.py', line 80 in __call__
return self.application(environ, start_response)
File '/data/galaxy-dev/galaxy-dev/lib/galaxy/web/framework/middleware/remoteuser.py', line 91 in __call__
return self.app( environ, start_response )
File '/data/galaxy-dev/galaxy-dev/eggs/Paste-1.6-py2.6.egg/paste/httpexceptions.py', line 632 in __call__
return self.application(environ, start_response)
File '/data/galaxy-dev/galaxy-dev/lib/galaxy/web/framework/base.py', line 160 in __call__
body = method( trans, **kwargs )
File '/data/galaxy-dev/galaxy-dev/lib/galaxy/web/framework/__init__.py', line 184 in decorator
return func( self, trans, *args, **kwargs )
File '/data/galaxy-dev/galaxy-dev/lib/galaxy/web/controllers/admin_toolshed.py', line 1469 in update_to_changeset_revision
update_repository( repo, latest_ctx_rev )
File '/data/galaxy-dev/galaxy-dev/lib/galaxy/util/shed_util.py', line 1655 in update_repository
rev=ctx_rev )
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/commands.py', line 5661 in update
rev = scmutil.revsingle(repo, rev, rev).rev()
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/scmutil.py', line 515 in revsingle
l = revrange(repo, [revspec])
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/scmutil.py', line 588 in revrange
for r in m(repo, range(len(repo))):
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/revset.py', line 1414 in mfunc
return getset(repo, subset, tree)
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/revset.py', line 156 in getset
return methods[x[0]](repo, subset, *x[1:])
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/revset.py', line 171 in symbolset
return stringset(repo, subset, x)
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/revset.py', line 161 in stringset
x = repo[x].rev()
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/localrepo.py', line 235 in __getitem__
return context.changectx(self, changeid)
File '/data/galaxy-dev/galaxy-dev/eggs/mercurial-2.2.3-py2.6-linux-x86_64-ucs2.egg/mercurial/context.py', line 103 in __init__
_("unknown revision '%s'") % changeid)
RepoLookupError: unknown revision '4'


Lance Parsons wrote:

Thanks Greg. I used you're updated version and added numpy as a
separate dependency. It seems to work in my development system. I
haven't updated my qa or production versions yet, so I can't check
there. Perhaps you could test it and let me know if there are issues, etc.

Glad I can be of help.

Lance

Greg Von Kuster wrote:

Hello Lance,

See my inline comments.


On Sep 4, 2012, at 3:15 PM, Lance Parsons wrote:


I've put together a tool wrapper for the htseq-count script that is part of the HTSeq python package and uploaded that to the tool shed.  However, I have discovered that the tool dependancies do not install properly.  There are a couple of issues that I've run into.

1) The biggest issue is the inability to set both the PYTHONPATH and PATH environment variables as part of installation.  If I put two separate<action type="set_environment">  tags, then the second overwrites the first in the env.sh file. If instead, I put two<environment_variable>  tags within the action, only one (the second) gets executed during installation.


The above issues have been corrected in changeset 7621:108cda898646, which is currently available only in the Galaxy central repository.  There is a new Galaxy distribution scheduled to go out today, but unfortunately this changeset will not be included.  From here on, Galaxy releases are tentatively scheduled for every 2 weeks, so if you don't want to pull from the Galaxy central repository, the fix should be available in the next release 2 weeks from now.

I've slightly altered your tool_dependencies.xml file (mostly because I've implemented support for a new "make_directory" tag.  It is attached below - you should updated your tool_dependencies.xml file in your htseq_count repository with this updated version.



2) Another issue is the inability to install Numpy as a dependancy to HTSeq.  I can get Numpy to install, but it won't be in the PYTHONPATH when HTSeq is installed (or executed).


Were you attempting to install numpy as another package defined in your tool-dependencies.xml file?   If so, the fixes in the above changeset should allow you to do that now since you can install it before htseq and set PYTHONPATH to point to numpy  as part of the installation.

Let me know if you need additional help or bump into further problems.   The tool dependency installation components are fairly immature, so adding enhancements for new tools like yours is very valuable.

Thanks for your help on this,

Greg Von Kuster



Any thoughts on how to address these issues would be appreciated.

--
Lance Parsons - Scientific Programmer
134 Carl C. Icahn Laboratory
Lewis-Sigler Institute for Integrative Genomics
Princeton University

___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:

http://lists.bx.psu.edu/




--
Lance Parsons - Scientific Programmer
134 Carl C. Icahn Laboratory
Lewis-Sigler Institute for Integrative Genomics
Princeton University

___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:

   http://lists.bx.psu.edu/


--
Lance Parsons - Scientific Programmer
134 Carl C. Icahn Laboratory
Lewis-Sigler Institute for Integrative Genomics
Princeton University



--
Lance Parsons - Scientific Programmer
134 Carl C. Icahn Laboratory
Lewis-Sigler Institute for Integrative Genomics
Princeton University

___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:

  http://lists.bx.psu.edu/

--
Lance Parsons - Scientific Programmer
134 Carl C. Icahn Laboratory
Lewis-Sigler Institute for Integrative Genomics
Princeton University



--
Lance Parsons - Scientific Programmer
134 Carl C. Icahn Laboratory
Lewis-Sigler Institute for Integrative Genomics
Princeton University