Q: How can a Rake task know the caller's directory?

J

James Britt

I'm increasingly using rake files for common system tasks. Since Rake
will search up a directory tree until it finds a Rakefile, I can have
one in my home directory and call tasks from anyplace below that.

But tasks that need to know something about the current directory have
me stymied. Calling Dir.pwd in a task reflects the directory of the
Rakefile.

Is there a rake method that returns the path of where the Rakefile
search began?


E.g., if I'm in /home/james/foo/bar, and the nearest Rakefile is in
/home/james, with task :whereami, and I call

$ rake whereami

what does :whereami have to do to return '/home/james/foo/bar' ?

Thanks!
 
J

Joel VanderWerf

James said:
I'm increasingly using rake files for common system tasks. Since Rake
will search up a directory tree until it finds a Rakefile, I can have
one in my home directory and call tasks from anyplace below that.

But tasks that need to know something about the current directory have
me stymied. Calling Dir.pwd in a task reflects the directory of the
Rakefile.

Is there a rake method that returns the path of where the Rakefile
search began?


E.g., if I'm in /home/james/foo/bar, and the nearest Rakefile is in
/home/james, with task :whereami, and I call

$ rake whereami

what does :whereami have to do to return '/home/james/foo/bar' ?

No idea. One possible solution is to put a "stub" rakefile in the dir,
and the stub requires the main rakefile. Then (IIRC) the pwd during the
rake tasks will be that of the stub. But maybe you don't want to scatter
stubs all over the place.
 
M

Mike Harris

James said:
I'm increasingly using rake files for common system tasks. Since Rake
will search up a directory tree until it finds a Rakefile, I can have
one in my home directory and call tasks from anyplace below that.

But tasks that need to know something about the current directory have
me stymied. Calling Dir.pwd in a task reflects the directory of the
Rakefile.

Is there a rake method that returns the path of where the Rakefile
search began?


E.g., if I'm in /home/james/foo/bar, and the nearest Rakefile is in
/home/james, with task :whereami, and I call

$ rake whereami

what does :whereami have to do to return '/home/james/foo/bar' ?

Thanks!
Modify rake to log the initial working directory before doing anything
else. You could do this by putting the line to log the dir in the file
bin/rake. Take advantage of ruby's openness.

Alternatively, you could put a file in a dir on your PATH (let's say
rake2) to log the dir and then delegate to rake, and just call rake2
instead of rake from the command line.
 
J

James Britt

Mike said:
Modify rake to log the initial working directory before doing anything
else. You could do this by putting the line to log the dir in the file
bin/rake. Take advantage of ruby's openness.

I've hacked ruby apps in this manner before, but they are prone to "gem
update <name_of_gem>" stomping.

So any mods to rake itself have to happen via by a plugin that will
survive library updates.
Alternatively, you could put a file in a dir on your PATH (let's say
rake2) to log the dir and then delegate to rake, and just call rake2
instead of rake from the command line.

Too hackish, even for me. I'd rather see about a cleaner way for any
arbitrary rake task to know the initial calling directory.

--
James Britt

"I have the uncomfortable feeling that others are making a religion
out of it, as if the conceptual problems of programming could be
solved by a single trick, by a simple form of coding discipline!"
- Edsger Dijkstra
 
J

James Britt

Joel said:
No idea. One possible solution is to put a "stub" rakefile in the dir,
and the stub requires the main rakefile. Then (IIRC) the pwd during the
rake tasks will be that of the stub. But maybe you don't want to scatter
stubs all over the place.

Right, that's the whole point. For example, I have a Rake task that
will do an svn commit. I've also aliased rake to r , so I can do this
to commit code:

$ r com This is my log message

and the :com task knows how to grab the message, do the commit, and run
a few other things (like svn status to alert if I missed adding any new
files).

Very handy, except as at stands, the invoked rake file thinks everything
happens relative where it lives, so I have to create a Rakefile
everyplace I want to use a location-Dependant task.


(Someone else mentioned using a "rake2" delegate; since I'm already
using 'r' as my rake invocation, maybe having a real r.rb that stores
the calling directory in ENV might work. That rake tasks that need the
caller dir can look for the value in ENV; these tasks could then assume
a default dir if this environment variable is empty.)

--
James Britt

"I have the uncomfortable feeling that others are making a religion
out of it, as if the conceptual problems of programming could be
solved by a single trick, by a simple form of coding discipline!"
- Edsger Dijkstra
 
J

James Britt

Mike said:
Alternatively, you could put a file in a dir on your PATH (let's say
rake2) to log the dir and then delegate to rake, and just call rake2
instead of rake from the command line.

I've tried that out, and it works quite well.

I had already aliased 'r' to 'rake' to save tying those extra three
characters. I now have a real script 'r' (well, "r.rb", which is now
the new r alias):


#!/usr/local/bin/ruby
ENV['cwd'] = Dir.pwd
puts `rake #{ARGV.join( ' ')}`

Tasks that need the caller's working directory now use

ENV['cwd'] || Dir.pwd



Thanks!
 
T

Trans

James said:
Right, that's the whole point. For example, I have a Rake task that
will do an svn commit. I've also aliased rake to r , so I can do this
to commit code:

$ r com This is my log message

how are you accessing "This is my log message" in your task?

t.
 
J

James Britt

Trans said:
how are you accessing "This is my log message" in your task?

ARGV.shift # remove the task name
ARGV.join( ' ') # Get everything else on the command line

Sadly, rake still thinks that there are more tasks on the command line;
even calling ARGV.clear doesn't help.

(I looked for the command to flush the Rake task call queue, but
couldn't find it, so I live with the error it raises at the end. :( )



--
James Britt

http://www.ruby-doc.org - Ruby Help & Documentation
http://www.artima.com/rubycs/ - The Journal By & For Rubyists
http://www.rubystuff.com - The Ruby Store for Ruby Stuff
 
J

Jim Weirich

James said:
I'm increasingly using rake files for common system tasks. Since Rake
will search up a directory tree until it finds a Rakefile, I can have
one in my home directory and call tasks from anyplace below that.

But tasks that need to know something about the current directory have
me stymied. Calling Dir.pwd in a task reflects the directory of the
Rakefile.

Is there a rake method that returns the path of where the Rakefile
search began?


E.g., if I'm in /home/james/foo/bar, and the nearest Rakefile is in
/home/james, with task :whereami, and I call

$ rake whereami

what does :whereami have to do to return '/home/james/foo/bar' ?

Thanks!

task :whereami do
puts Rake.original_dir
end


-- Jim Weirich
 
J

Jim Weirich

James said:
ARGV.shift # remove the task name
ARGV.join( ' ') # Get everything else on the command line

Sadly, rake still thinks that there are more tasks on the command line;
even calling ARGV.clear doesn't help.

(I looked for the command to flush the Rake task call queue, but
couldn't find it, so I live with the error it raises at the end. :( )

You could do:

rake com MSG="This is my log message"

Inside of your Rakefile, ENV['MSG'] will contain your log message.

Sadly, more typing that what you have, but it works with rake instead of
against it.

-- Jim Weirich
 
T

Trans

Jim said:
James said:
ARGV.shift # remove the task name
ARGV.join( ' ') # Get everything else on the command line

Sadly, rake still thinks that there are more tasks on the command line;
even calling ARGV.clear doesn't help.

(I looked for the command to flush the Rake task call queue, but
couldn't find it, so I live with the error it raises at the end. :( )

You could do:

rake com MSG="This is my log message"

Inside of your Rakefile, ENV['MSG'] will contain your log message.

Sadly, more typing that what you have, but it works with rake instead of
against it.

why did you choose to use environment vars here. isn't that sort of
like using globals? also, these don't work in my case becuase i can't
differentiate the env vars set by rake from the "real" env vars.

t.
 
M

Mike Harris

task :whereami do
puts Rake.original_dir
end


-- Jim Weirich

Checkmate, Mr. Weirich. I looked in run/load_rakefile, but neglected to
look in initialize (or the changelog). Thanks.
 
J

James Britt

Jim said:
You could do:

rake com MSG="This is my log message"

Inside of your Rakefile, ENV['MSG'] will contain your log message.

Right, but too much effort for the user. Leaves too much room for James
to forget a quote or mistype the var name.

Sadly, more typing that what you have, but it works with rake instead of
against it.

But isn't there a way to tell Rake to ignore any additional tasks it
thinks are queued up?

When I run

$rake com This is my message

it works fine, but at the very end rake complains that it doesn't know
anything about task 'This'.


Thanks,
 
J

Jim Weirich

Trans said:
why did you choose to use environment vars here. isn't that sort of
like using globals? also, these don't work in my case becuase i can't
differentiate the env vars set by rake from the "real" env vars.

The fact they look like environment variables is delibrate. This allows
you to pass information to Rake via environment variables or by command
line and Rake doesn't care. You can put a default value in your
environment and then easily override it on the command line. It is also
the way that the original make program worked.

-- Jim Weirich
 
J

Jim Weirich

Mike said:
Checkmate, Mr. Weirich. I looked in run/load_rakefile, but neglected to
look in initialize (or the changelog). Thanks.

You can thank Jim Freeze for suggesting this feature.

-- Jim Weirich
 
J

James Edward Gray II

Jim said:
James said:
Trans wrote:
$ r com This is my log message

how are you accessing "This is my log message" in your task?

ARGV.shift # remove the task name
ARGV.join( ' ') # Get everything else on the command line

Sadly, rake still thinks that there are more tasks on the command
line;
even calling ARGV.clear doesn't help.

(I looked for the command to flush the Rake task call queue, but
couldn't find it, so I live with the error it raises at the end. :
( )

You could do:

rake com MSG="This is my log message"

Inside of your Rakefile, ENV['MSG'] will contain your log message.

Sadly, more typing that what you have, but it works with rake
instead of
against it.

why did you choose to use environment vars here. isn't that sort of
like using globals? also, these don't work in my case becuase i can't
differentiate the env vars set by rake from the "real" env vars.

Sure you can. Just introduce a little pragmatic namespacing:

rake com RAKE_MSG="This is my log message"

James Edward Gray II
 
J

James Britt

Jim said:
task :whereami do
puts Rake.original_dir
end

I had looked for such a method in ri and the Rake rdocs, but saw nothing
that looked like what I wanted.

$ ri Rake
------------------------------------------------------------ Class: Rake
(no description...)
------------------------------------------------------------------------


I could not find the originaL_dir method listed in the rdoc either,
until I walked through the source code.

Because of the way this method is defined it Rake, rdoc will not
correctly process it as a class method on Rake itself:

# in lib/rake.rb

#
class << self
# Current Rake Application
def application
@application ||= Rake::Application.new
end

# Set the current Rake application object.
def application=(app)
@application = app
end

# Return the original directory where the Rake application was
# started.
def original_dir
application.original_dir
end

end


And in the rdoc for Rake::Application, the method is not listed in the
list of methods because it gets rdoc'ed as an "attribute".

One has to already know where to look, and what to look for, to find the
behavior.
 
T

Trans

James said:
Sure you can. Just introduce a little pragmatic namespacing:

rake com RAKE_MSG="This is my log message"

Hmm... maybe. Think I'll just use a single env var and parse that I
guess:

rake com RAKEOPTS="This is my log message"

in my case

rake sign RAKEOPTS="--digest=sha256"

personally, it would cooler if Rake took some of Ara's kung-fu and used
someting like a - to separate task from options.

T.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

Forum statistics

Threads
473,744
Messages
2,569,483
Members
44,901
Latest member
Noble71S45

Latest Threads

Top