ANN: doit 0.2.0 released
doit comes from the idea of bringing the power of build-tools to execute any kind of task. It will keep track of dependencies between “tasks” and execute them only when necessary. It was designed to be easy to use and “get out of your way”. check the new website http://python-doit.sourceforge.net/ -- http://mail.python.org/mailman/listinfo/python-list
[ANN] DoIt 0.1.0 Released (build tool)
DoIt - A task execution tool (build-tool) = This is the first public release of DoIt Website: http://python-doit.sourceforge.net/ Release: DoIt 0.1.0 License: MIT About - DoIt is a build tool that focus not only on making/building things but on executing any kind of tasks in an efficient way. Designed to be easy to use and "get out of your way". DoIt like most build tools is used to execute tasks defined in a configuration file. Configuration files are python modules. The tasks can be python functions (or any callable) or an external shell script. DoIt automatically keeps track of declared dependencies executing only tasks that needs to be update (based on which dependencies have changed). In DoIt, unlike most(all?) build-tools, a task doesn't need to define a target file to use the execute only if not up-to-date feature. This make DoIt specially suitable for running test suites. DoIt can be used to perform any task or build anything, though it doesn't support automatic dependency discovery for any language. Cheers, Eduardo -- http://mail.python.org/mailman/listinfo/python-list
Re: random.random(), random not defined!?
On Sun, Apr 20, 2008 at 12:58 AM, globalrev <[EMAIL PROTECTED]> wrote: > do i need to import something to use random? > -- you need to import random :) Python 2.5.1 (r251:54863, Mar 7 2008, 03:39:23) [GCC 4.1.3 20070929 (prerelease) (Ubuntu 4.1.2-16ubuntu2)] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import random >>> random.random() 0.76018998919085967 -- http://mail.python.org/mailman/listinfo/python-list
Re: [ANN] DoIt 0.1.0 Released (build tool)
On Sun, Apr 20, 2008 at 2:04 AM, John Machin <[EMAIL PROTECTED]> wrote:
> You may like to consider the possibility of confusion caused by the
> similarity of some characters in some fonts (DoIt, Do1t, Dolt) ...
> google("dictionary dolt") :-)
> --
> http://mail.python.org/mailman/listinfo/python-list
>
hehehehe. i feel like a DOLT
I never realized that it is confusing. I also didnt know this word
"dolt" before.
i wont use capital letters anymore.
thanks
--
http://mail.python.org/mailman/listinfo/python-list
Re: Python "make" like tools (was Re: [ANN] DoIt 0.1.0 Released (build tool))
>
> I took a look at dolt syntax, and saw this:
>
> QQQ
>
> def create_folder(path):
> """Create folder given by "path" if it doesnt exist"""
> if not os.path.exists(path):
> os.mkdir(path)
> return True
>
> def task_create_build_folder():
> buildFolder = jsPath + "build"
> return {'action':create_folder,
> 'args': (buildFolder,)
> }
>
> QQQ
>
> Wouldn't it be more convenient to provide syntax like this:
>
> @task("create_build_folder")
> @depend("dep1 some_other_dep")
> def buildf():
> buildFolder = jsPath + "build"
> create_folder(buildFolder)
>
> I find the doit syntax a bit cumbersome, especially as you can avoid
> 'args' by just returning a lamda in 'action'.
My idea was to: do *not* add any new syntax (to avoid being
cumbersome). It is just python, you dont have to import or subclass
anything. You just need to create a function that returns a dictionary
with some predefined keys.
I though about using decorators in the beginning... but returning a
dictionary looked easier to implement and more flexible. one important
feature is how easy to define a group of task with the same action.
Take a look at the example below on running pychecker in all python
files from a folder. I couldnt figure out an easy way of doing it with
decorators.
import glob;
pyFiles = glob.glob('*.py')
def task_checker():
for f in pyFiles:
yield {'action': "pychecker %s"% f,
'name':f,
'dependencies':(f,)}
Another advantage of using just a dictionary to define a task is that
it will be easy to read tasks from a text file (if it is very simple
and you dont need to write any python script). but not implemented
yet.
I though about using decorator for simple python-tasks but in a different way:
@task
def create_folder(path):
"""Create folder given by "path" if it doesnt exist"""
if not os.path.exists(path):
os.mkdir(path)
return True
so if your python function is a task and you will use it only once you
dont need to define a function for the 'action' and another function
to create the task. but not implement yet also.
>
> I've looked around a bit for python "make" replacement, but there does
> not seem to be a simple & straightforward solution around (read -
> straight-python syntax, one .py file installation, friendly license).
apart from one .py file installation (easy_install is not enough?)
thats what i am trying to do.
thanks for the feedback.
cheers,
Eduardo
--
http://mail.python.org/mailman/listinfo/python-list
Re: Python "make" like tools (was Re: [ANN] DoIt 0.1.0 Released (build tool))
I guess I should post a link to the project in this thread... http://python-doit.sourceforge.net/ -- http://mail.python.org/mailman/listinfo/python-list
Re: Python "make" like tools (was Re: [ANN] DoIt 0.1.0 Released (build tool))
On Wed, Apr 23, 2008 at 8:39 PM, Ville M. Vainio <[EMAIL PROTECTED]> wrote:
>
> Yeah, decorators get around this.
>
>
> Perhaps you could do:
>
> for f in pyFiles:
> @task("checker")
> @depend(f)
> def check():
> c("pychecker %s" % f)
>
> Never underestimate the magic that is nested scopes and name-agnostic
> function object creation...
>
ok. it is possible to do everything with decorators... i agree.
but i dont want "name-agnostic function object creation". i want to be
able to execute every single task without executing all other tasks.
thats why when defining sub-tasks it is required an extra parameter
"name". I know you could add another decorator for this also :)
I mean decorators could do the job i dont say they cant. *I* prefer to
work with dictionaries though. Even if they are a bit more verbose.
> >
> > I though about using decorator for simple python-tasks but in a different
> way:
> >
> > @task
> >
> > def create_folder(path):
> >"""Create folder given by "path" if it doesnt exist"""
> >if not os.path.exists(path):
> >os.mkdir(path)
> > return True
> >
> >
> > so if your python function is a task and you will use it only once you
> > dont need to define a function for the 'action' and another function
> > to create the task. but not implement yet also.
> >
>
> Yeah, this is what I consider much friendlier syntax (the aim is to not be
> much more verbose than make).
>
>
Maybe it is just a bad example. I tried to put all features on single
example. but in *real* life I could do like this.
import os
jsPath = "./"
jsFiles = ["file1.js", "file2.js"]
sourceFiles = [jsPath + f for f in jsFiles]
compressedFiles = [jsPath + "build/" + f + ".compressed" for f in jsFiles]
def task_shrink_js():
# create output folder
if not os.path.exists(path):
os.mkdir(path)
for jsFile,compFile in zip(sourceFiles,compressedFiles):
action = 'java -jar custom_rhino.jar -c %s > %s'% (jsFile, compFile)
yield {'action':action,
'name':jsFile,
'dependencies':(jsFile,),
'targets':(compFile,)
}
Can I challenge you to do this with any other build tool?
i guess anything more complicated than this you want to create a
python function separate from the task definition anyway.
My examples are from scratch. I dont use a "standard library".
I could also document the short-cuts :P. I was afraid that including
the short-cuts would make it more complicated for who is trying to
learn it.
if you dont define any dependencies or args to the function you dont
need to return a dictionary. just the action:
def task_simple_shell():
return "echo spam"
def task_simple_python():
def say_spam():
print "spam"
return True
return say_spam
# or from our discussion
def task_create_build_folder():
def create_folder():
path = jsPath + "build"
if not os.path.exists(path):
os.mkdir(path)
return True
return create_folder
>
> > apart from one .py file installation (easy_install is not enough?)
> > thats what i am trying to do.
> >
>
> easy_install is not really enough - it introduces a dependency that you
> can't get around by just shipping a short .py file with your project.
>
what about shipping a single egg file with the whole package? ( I am
not very much familiar with this subject)
> 'Paver' seems to have the right idea:
>
> http://www.blueskyonmars.com/projects/paver/index.html
>
Thanks for the link. I took a quick look on its documentation.
It seems that "paver" doesnt keep track of file-dependencies at all.
So it is more like a setuptools extension commands than a
"full-feature" build system with "smart" re-build and dependency
support.
"doit" target is not on creating releases for python projects only. it
can do it also, but it can do much more.
You can read about my motivation to start another build tool project
on http://schettino72.wordpress.com/2008/04/14/doit-a-build-tool-tale/
> But it's still *slightly* too big:
>
man, it is hard to make you happy :)
the doit egg file containg the whole packge is 27782 bytes on my
system. but you also need the command line script 154 bytes.
cheers,
Eduardo
--
http://mail.python.org/mailman/listinfo/python-list
[ANN] doit 0.5
doit - Automation Tool doit comes from the idea of bringing the power of build-tools to execute any kind of task. It will keep track of dependencies between "tasks" and execute them only when necessary. It was designed to be easy to use and "get out of your way". doit can be used as: * a build tool (generic and flexible) * home of your management scripts (it helps you organize and combine shell scripts and python scripts) * a functional tests runner (combine together different tools) homepage: http://python-doit.sourceforge.net/ PyPi: http://pypi.python.org/pypi/doit license: MIT contact: https://launchpad.net/~schettino72 Regards, Eduardo -- http://mail.python.org/mailman/listinfo/python-list
Re: Wanted: Python solution for ordering dependencies
On Sun, Apr 25, 2010 at 4:53 AM, Jonathan Fine wrote: > Hi > > I'm hoping to avoid reinventing a wheel (or other rolling device). I've got > a number of dependencies and, if possible, I want to order them so that each > item has its dependencies met before it is processed. > > I think I could get what I want by writing and running a suitable makefile, > but that seems to be such a kludge. > > Does anyone know of an easily available Python solution? http://pypi.python.org/pypi/doit -- http://mail.python.org/mailman/listinfo/python-list
Re: Wanted: Python solution for ordering dependencies
On Sun, Apr 25, 2010 at 11:44 PM, Jonathan Fine wrote:
> Eduardo Schettino wrote:
>>
>> On Sun, Apr 25, 2010 at 4:53 AM, Jonathan Fine wrote:
>>>
>>> Hi
>>>
>>> I'm hoping to avoid reinventing a wheel (or other rolling device). I've
>>> got
>>> a number of dependencies and, if possible, I want to order them so that
>>> each
>>> item has its dependencies met before it is processed.
>>>
>>> I think I could get what I want by writing and running a suitable
>>> makefile,
>>> but that seems to be such a kludge.
>>>
>>> Does anyone know of an easily available Python solution?
>>
>> http://pypi.python.org/pypi/doit
>
>
> Thank you for this, Eduardo. However, all I require is a means of ordering
> the items that respects the dependencies. This rest I can, and pretty much
> have to, manage myself.
>
> So probably something more lightweight would suit me.
you just want a function?
def order_tasks(tasks):
ADDING, ADDED = 0, 1
status = {} # key task-name, value: ADDING, ADDED
task_order = []
def add_task(task_name):
if task_name in status:
# check task was alaready added
if status[task_name] == ADDED:
return
# detect cyclic/recursive dependencies
if status[task_name] == ADDING:
msg = "Cyclic/recursive dependencies for task %s"
raise Exception(msg % task_name)
status[task_name] = ADDING
# add dependencies first
for dependency in tasks[task_name]:
add_task(dependency)
# add itself
task_order.append(task_name)
status[task_name] = ADDED
for name in tasks.keys():
add_task(name)
return task_order
if __name__ == '__main__':
task_list = {'a':['b','c'],
'b':['c'],
'c':[]}
print order_tasks(task_list)
--
http://mail.python.org/mailman/listinfo/python-list
[ANN] doit - automation tool 0.8 released
pypi: http://pypi.python.org/pypi/doit homepage: http://python-doit.sourceforge.net/ `doit` comes from the idea of bringing the power of build-tools to execute any kind of task. It will keep track of dependencies between "tasks" and execute them only when necessary. It was designed to be easy to use and "get out of your way". Features: * Easy to use, "no-API" * Use python to dynamically create tasks on-the-fly * Flexible, adapts to many workflows for creation of tasks/rules/recipes * Support for multi-process parallel execution * Built-in integration of inotify (automatically re-execution) `doit` can be used as: * a build tool (generic and flexible) * home of your management scripts (it helps you organize and combine shell scripts and python scripts) * a functional tests runner (combine together different tools) -- http://mail.python.org/mailman/listinfo/python-list
