[jira] [Commented] (AIRAVATA-3032) experiment data directory should be fixed and unique

["ASF subversion and git services (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/AIRAVATA-3032?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16849717#comment-16849717 ] 

ASF subversion and git services commented on AIRAVATA-3032:
-----------------------------------------------------------

Commit 8657f9d775c8debcb579fbdfe78e9fa293011616 in airavata-django-portal's branch refs/heads/airavata-3016 from Marcus Christie
[ https://gitbox.apache.org/repos/asf?p=airavata-django-portal.git;h=8657f9d ]

AIRAVATA-3032 Add isInputFileUpload to DataProductSerializer

Use isInputFileUpload to distinguish between input files that were
uploaded from ones that are selected from user storage. This is
then used to delete input files that were uploaded when the user no
longer wants to use that file in an experiment.


> experiment data directory should be fixed and unique
> ----------------------------------------------------
>
>                 Key: AIRAVATA-3032
>                 URL: https://issues.apache.org/jira/browse/AIRAVATA-3032
>             Project: Airavata
>          Issue Type: Sub-task
>          Components: Django Portal
>            Reporter: Marcus Christie
>            Assignee: Marcus Christie
>            Priority: Major
>             Fix For: 0.19
>
>
> The experiment data directory should use a timestamp or some other unique bits added to the experiment name to make it unique. It should also be fixed once set and not change when the experiment name changes.
> h5. Alternate Design
> The problem with the current approach is that uploaded input files are saved into an experiment data directory following the {{<project name>/<experiment name>}} pattern, but experiment name may not be unique. But also, the experiment name can be changed until launch time. So to fix the one problem I need to fix the other one and my initial thought was to just set the experiment data dir to {{<project name>/<experiment name>}} at the first file upload. However that has a couple problems: 1) the name of the experiment data dir may not match the name of the experiment, 2) it gets a little complicated in the UI to figure out a unique experiment data dir and pass that down from the ExperimentEditor to the FileInputEditors.
> So I've come up with an alternate approach that involves saving uploaded input files into the user's tmp directory and moving them from this tmp directory to the experiment data dir at launch time. This simplifies the upload process and it allows creating an experiment data dir with the name of the experiment as provided by the user at *Save and Launch* time.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)