Hey Rob,

I haven't touched in quite some time, but from the comments it indicates
that the sleep was because the datatype hadn't been detected and set yet,
so that's what I'd check for -- specifically result_ds['data_type'] being
something sensible (I have no idea what it is by default off the top of my
head).

Let me know if this doesn't work and I can take a closer look.

-Dannon


On Mon, Apr 29, 2013 at 7:37 PM, Rob Leclerc <robert.lecl...@gmail.com>wrote:

> Correction. The above were not reliable methods to ensure the file was
> copied into the data library. Checking for file_size != 0 was also not
> effective for large files.
>
> Dannon, can you tell me which field we should query and what state/message
> which will allow us to avoid race conditions?
>
> The only solution that I can see that when the file_size != 0, you then
> ensure it that the file_size has not changed after a a short delay.
>
> Cheers,
> Rob
>
>
>
> Rob Leclerc, PhD
> <http://www.linkedin.com/in/robleclerc><https://twitter.com/#!/robleclerc>
> P: (US) +1-(917)-873-3037
> P: (Shanghai) +86-1-(861)-612-5469
> Personal Email: rob.lecl...@aya.yale.edu
>
>
> On Mon, Apr 29, 2013 at 4:11 PM, Rob Leclerc <robert.lecl...@gmail.com>wrote:
>
>> Hi Dannon,
>>
>> I've written some code to (i) query a dataset to ensure that it's been
>> uploaded after a submit and (ii) to ensure a resulting dataset has been
>> written to the file.
>>
>> *#Block until all datasets have been uploaded*
>> libset = submit(api_key, api_url + "libraries/%s/contents" % library_id,
>> data, return_formatted = False)
>> for ds in libset:
>>     while True:
>>         uploaded_file = display(api_key, api_url +
>> 'libraries/%s/contents/%s' %(library_id, ds['id']), return_formatted=False)
>>         if uploaded_file['misc_info'] == None:
>>             time.sleep(1)
>>         else:
>>             break
>>
>> *#Block until all result datasets have been saved to the filesystem*
>> result_ds_url = api_url + 'histories/' + history_id + '/contents/' +
>> dsh['id'];
>> while True:
>>     result_ds = display(api_key, result_ds_url, return_formatted=False)
>>         if result_ds["state"] == 'ok':
>>             break
>>         else:
>>             time.sleep(1)
>>
>>
>> Rob Leclerc, PhD
>> <http://www.linkedin.com/in/robleclerc><https://twitter.com/#!/robleclerc>
>> P: (US) +1-(917)-873-3037
>> P: (Shanghai) +86-1-(861)-612-5469
>> Personal Email: rob.lecl...@aya.yale.edu
>>
>>
>> On Mon, Apr 29, 2013 at 11:18 AM, Dannon Baker <dannon.ba...@gmail.com>wrote:
>>
>>> Yep, that example filesystem_paths you suggest should work fine.  The
>>> sleep() bit was a complete hack from the start, for simplicity in
>>> demonstrating a very basic pipeline, but what you probably want to do for a
>>> real implementation is query the dataset in question via the API, verify
>>> that the datatype/etc have been set, and only after that execute the
>>> workflow; instead of relying on sleep.
>>>
>>>
>>> On Mon, Apr 29, 2013 at 9:24 AM, Rob Leclerc 
>>> <robert.lecl...@gmail.com>wrote:
>>>
>>>> Hi Dannon,
>>>>
>>>> Thanks for the response. Sorry to be pedantic, but just to make sure
>>>> that I understand the interpretation of this field on the other side of the
>>>> API, I would need to have something like the following:
>>>>
>>>> data['filesystem_paths'] = "/home/me/file1.vcf \n /home/me/file2.vcf /n
>>>> /home/me/file3.vcf"
>>>>
>>>> I assume I should also increase the time.sleep() to reflect the
>>>> uploading of extra files?
>>>>
>>>> Cheers,
>>>>
>>>> Rob
>>>>
>>>> Rob Leclerc, PhD
>>>> <http://www.linkedin.com/in/robleclerc><https://twitter.com/#!/robleclerc>
>>>> P: (US) +1-(917)-873-3037
>>>> P: (Shanghai) +86-1-(861)-612-5469
>>>> Personal Email: rob.lecl...@aya.yale.edu
>>>>
>>>>
>>>> On Mon, Apr 29, 2013 at 9:15 AM, Dannon Baker 
>>>> <dannon.ba...@gmail.com>wrote:
>>>>
>>>>> Hey Rob,
>>>>>
>>>>> That example_watch_folder.py does just submit exactly one at a time,
>>>>> executes the workflow, and then does the next all in separate 
>>>>> transactions.
>>>>>  If you wanted to upload multiple filepaths at once, you'd just append 
>>>>> more
>>>>> to the ''filesystem_paths' field (newline separated paths).
>>>>>
>>>>> -Dannon
>>>>>
>>>>>
>>>>> On Fri, Apr 26, 2013 at 11:54 PM, Rob Leclerc <
>>>>> robert.lecl...@gmail.com> wrote:
>>>>>
>>>>>> I'm looking at example_watch_folder.py and it's not clear from the
>>>>>> example how you submit multiple datasets to a library. In the example, 
>>>>>> the
>>>>>> first submit returns a libset [] with only a single entry and then 
>>>>>> proceeds
>>>>>> to iterate through each dataset in the libset in the following section:
>>>>>>
>>>>>> data = {}
>>>>>>
>>>>>>    data['folder_id'] = library_folder_id
>>>>>>
>>>>>>    data['file_type'] = 'auto'
>>>>>>
>>>>>>    data['dbkey'] = ''
>>>>>>
>>>>>>    data['upload_option'] = 'upload_paths'
>>>>>>
>>>>>>
>>>>>>
>>>>>> *data['filesystem_paths'] = fullpath*
>>>>>>
>>>>>>    data['create_type'] = 'file'
>>>>>>
>>>>>>    libset = submit(api_key, api_url + "libraries/%s/contents" %
>>>>>> library_id, data, return_formatted = False)
>>>>>>
>>>>>>    time.sleep(5)
>>>>>>
>>>>>>    for ds in libset:
>>>>>>
>>>>>>        if 'id' in ds:
>>>>>>
>>>>>>                         wf_data = {}
>>>>>>
>>>>>>                         wf_data['workflow_id'] = workflow['id']
>>>>>>
>>>>>>                         wf_data['history'] = "%s - %s" % (fname,
>>>>>> workflow['name'])
>>>>>>
>>>>>>                         wf_data['ds_map'] = {}
>>>>>>
>>>>>>                         for step_id, ds_in in workflow['inputs'
>>>>>> ].iteritems():
>>>>>>
>>>>>>                             wf_data['ds_map'][step_id] = {'src':'ld',
>>>>>> 'id':ds['id']}
>>>>>>
>>>>>>                         res = submit( api_key, api_url + 'workflows',
>>>>>> wf_data, return_formatted=False)
>>>>>>
>>>>>>
>>>>>>
>>>>>> Rob Leclerc, PhD
>>>>>> <http://www.linkedin.com/in/robleclerc><https://twitter.com/#!/robleclerc>
>>>>>> P: (US) +1-(917)-873-3037
>>>>>> P: (Shanghai) +86-1-(861)-612-5469
>>>>>> Personal Email: rob.lecl...@aya.yale.edu
>>>>>>
>>>>>> ___________________________________________________________
>>>>>> 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/
>>>>>>
>>>>>> To search Galaxy mailing lists use the unified search at:
>>>>>>   http://galaxyproject.org/search/mailinglists/
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>
___________________________________________________________
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/

To search Galaxy mailing lists use the unified search at:
  http://galaxyproject.org/search/mailinglists/

Reply via email to