[ome-users] Handling public data

[Josh Moore]

Hi all,

On Nov 16, 2014, at 11:09 PM, Jake Carroll wrote:

> Thanks for your replies, both of you.
> 
> The amazing thing about OMERO is that we oftentimes experience better time to feedback and responses (and quickly resolved problems) than we would get from companies we pay substantial sums of money to yearly, for service contracts. Weird how that works...

:)


> Anyway.
> 
> Thanks for the feedback. From my perspective it seems to make more sense that if a user wanted to put data into a public container or project such that members of the public could see it without being "authed" or, simply being authed with a public set of credentials, it makes more sense to "single instance" the object by derivation or a link to that set of inodes on a filesystem sitting inside the omeroFS directory structures.
> 
> Simple terms: let's symlink with some safety and sanity checks in place. Ergo all the public user can do is read it. They can't do anything else.

That's an interesting proposal. We've been calling the full-blown version of this "deep copy", meaning the metadata is copied but continues to use the same files on disk. A key worry has been, for example, guaranteeing that a delete of the public copy can't propagate to the original version. Symlinks would likely prevent the delete error (though here Java is less than ideal) but wouldn't necessarily prevent a modification (and Java is even *less* ideal for working with read/write permissions at the OS level).


> I'm probably missing something fundamental in the way OMERO currently works (and this is why you fellows haven't done what I've mentioned above yet!).
> 
> Else...nobody had ever asked ;).

Um, no.  :) The core burden is really producing a complete abstraction of all IO access so we can inject our permissions model at every level. Perhaps with further suggestions we can find a half-way house that will work for a large percentage of users in the short-term.


> Duplicating data isn't fun if we can avoid it. Equally, moving data out of know structures and ontologies that have been set up in an OMERO structure would be best left untouched if we could simply "flag" data as public in a set and as a result it would generate a public link to THAT SPECIFIC object.

This is largely how the "public share" functionality was intended to work.


> Happy to talk at length/white board it.
> Talk soon.

Sounds good.


> On 17 Nov 2014, at 08:01, Christophe TREFOIS wrote:
> 
> Why is it not possible to copy data to a public group ?
> 
> It does sound a little bit strange as users will have to break out their data from the structure they created.
> 
> Re-uploading is an option, but a little bit silly, as we would duplicate usage space.
> 
> Hopefully there will be a solution for that soon :)

:) At the end of the day (which it is), the issue is about handling a number of edge cases where data could be lost or corrupted. We've tried to err on the side of caution, though that certainly brings limitations.

Planned for the 5.1 series will be a number of improvements which simplify a general publication workflow: working with entire graphs, marking a group public, exporting more extensive data, etc. These tools will still require a duplication of the files, but perhaps with a combination of hard- and/or soft-links and *very careful permissions on your repository*, the disk usage won't be too burdensome.

As always, thanks for these & future suggestions!

Cheers,
~Josh.


> Best,
> 
> From: ome-users [mailto:ome-users-bounces at lists.openmicroscopy.org.uk] On Behalf Of William Moore
> Sent: dimanche, le 16 novembre 2014 22:58
> To: Jake Carroll
> Cc: ome-users at lists.openmicroscopy.org.uk<mailto:ome-users at lists.openmicroscopy.org.uk>
> Subject: Re: [ome-users] Handling public data
> 
> Hi Jake,
> 
> Any user on the OMERO that wants to make some of their data public needs to be added to the 'public group'.
> They can then move their data into that group. This will make it visible to the 'public user'.
> In OMERO.insight or web, you can right-click on an image in the left-panel hierarchy and choose "Move to group",
> then choose the "public" group. You can choose to organise images within Projects & Datasets in the
> public group if you wish, but it's not necessary for linking to the image via webgateway viewer.
> 
> Unfortunately it's not possible to copy/paste or automatically duplicate the image when you move
> them to the public group.
> If it's important to keep the original images in their current group, then you may want to
> re-import the images directly into the public group.
> 
> Apologies that a "one-click solution" isn't yet available, but hopefully the current workflow
> isn't too painful.
> 
> Hope that helps,
> 
>   Will.
> 
> 
> 
> On 16 Nov 2014, at 19:42, Jake Carroll <jake.carroll at uq.edu.au<mailto:jake.carroll at uq.edu.au>> wrote:
> 
> 
> Hi all,
> 
> Further to a question or two I had a week or so ago, I’ve been following this:
> 
> http://www.openmicroscopy.org/site/support/omero5/developers/Web/PublicData.html
> 
> I’m at the point where I’ve got a public user and a public group and I’ve got a string in the OMERO config such that:
> 
> omero.web.public.url_filter=/webgateway
> 
> But what next?
> 
> I can *see* the public user in the OMERO web client. It has Read-Only permissions set on the user/group.
> 
> So how does one then (as themselves or a “normal” user of OMERO and not the public-data user) effectively “copy” data into that user/group that lives in the public space?
> 
> An example.
> 
> * Researcher A wants some of the images he/she has created to be linked to in a paper. He/she wants a one-click solution in OMERO to provide this to a journal or paper.
> * The researcher already has plenty of data in OMERO sitting in his/her account under his/her own login and username, but that isn’t “public data”.
> 
> How or what steps in the OMERO GUI does a research take to “copy” their data (or pointers to it, depending on how omeroFS deals with these things…heh) so that it appears in the “public data” area such that it can be linked to like:
> 
> http://blah.super.cool.fqdn/webgateway/img_detail/IMAGE_ID/
> 
> To put it another way - how do they designate a piece of data to be “public” and then an appropriate or resulting public-data link is produced?
> 
> I thought I had it figured out, but when I realised as a normal user (as in, a logged in general protected user of OMERO) that they couldn’t simply “copy” and “paste” or “create project” in the public-data area, I realised this might require more thought. I’m effectively trying to help them create an environment where they themselves can designate data public with the click or two of a button and then simply paste a link - but…it doesn’t seem to be the case/or is that easy?
> 
> Or perhaps I’m missing something!
> 
> Talk soon and thank you for your time.
> 
> Regards,
> 
> -jc
>