Wikidata talk:WikiProject Informatics/FLOSS

From Wikidata
Jump to navigation Jump to search

Activate Wikidata:Flow[edit]

Tobias1984
Emw
Zuphilip
Danrok
Bene*
콩가루
TomT0m
DrSauron
Ruud Koot
Andreasburmeister
Ilya
Toto256
MichaelSchoenitzer
Metamorforme42
Pixeldomain
User:YULdigitalpreservation
Dipsode87
Pintoch
Daniel Mietchen
Jsamwrites
Tinker Bell
FabC
Jasc PL
putnik
Dhx1
Tris T7
Peb Aryan
lore.mazza004
Rc1959
Premeditated
Pictogram voting comment.svg Notified participants of WikiProject Informatics

Dachary
Metamorforme42
NMaia
Valerio Bozzolan
MichaelSchoenitzer (talk)
Jasc PL
LiberatorG
Pictogram voting comment.svg Notified participants of WikiProject Informatics/FLOSS

I propose that we activate Flow on this page. If there is no opposition within two weeks, I'll do it. Dachary (talk) 15:44, 16 July 2016 (UTC)

@Metamorforme42: I'm not sure how to do that on this page. Is it even possible ? – The preceding unsigned comment was added by Dachary (talk • contribs) at 15:44, 16 July 2016‎ (UTC).
@dachary: See Wikidata:Flow#Community page, Wikidata talk:Flow or eventually ask to Trizek (WMF).
ps: ping don't works if you forget signing.
Metamorforme42 (talk) 19:49, 24 August 2016 (UTC)

floss Catering isn’t a large contributor to open source[edit]

Unaffiliated developers are currently assigned to the organization floss.cc, a domain that happens to be owned and operated by bloss Catering in Germany. How about using an actual email address instead of picking on a random unaffiliated company? They must receive a bunch of email spam from Wikidata’s use of their domain. Propose changing contact@floss.cc to something at example.org or an organization that represents the interest of otherwise unaffiliated open source contributors (EFF? OSI?).

Mail threads[edit]

 – The preceding unsigned comment was added by Dachary (talk • contribs) at 09:17, 16 July 2016‎ (UTC).

Dual licensing[edit]

Liferay Portal (Q254211) actually are two slightly different software, one is distributed under a Free Software license and the other is under a proprietary license and has more features etc. I updated the license / instance of statements to add the proprietary aspects, with references so that it's less confusing. I however wonder if it would be better with two distinct items, related to each other. – The preceding unsigned comment was added by Dachary (talk • contribs) at 09:18, 16 July 2016 (UTC).

There is nothing wrong with having both. It is like having multiple editions of a book (with different ISBNs/product codes and licensing, prices, etc.). I think we can still consider them a single software. 50.53.1.33 00:29, 25 September 2016 (UTC)
When the only thing that changes is price and licensing, it makes sense to me. However the proprietary version of Liferay Portal (Q254211) is updated during a long period (4 years according to https://www.liferay.com/subscription-services Liferay Enterprise Subscription gives you four years of software updates per version. Liferay Community Edition is only updated once or twice after the first GA release) and becomes different from the free software version over time. This is an interesting case because there is only one name for the product although there are two different products, when you look close enough. When there are two different products and the developer gives them different names (such as GitLab EE (Q25973915) and GitLab (Q16639197)) and their code base is mostly the same, having two different items makes sense and there is no confusion. Dachary (talk) 07:22, 25 September 2016 (UTC)

For the record ISO Master (Q663640) appears to be a case of dual licensing where there is no difference between the proprietary software and the free software. Dachary (talk) 09:56, 28 September 2016 (UTC)

Identifiers[edit]

The Software directory entries and identifiers should be put together. I initially thought it would be best to have a chapter different from properties because it shows in a different part of the interface when editing an item. However, I'm happy to merge that with properties instead. @Metamorforme42: what do you think ? Dachary (talk) 08:23, 3 August 2016 (UTC)

You can merge. I just don't seen the section identifiers. Face-smile.svgMetamorforme42 (discussion) 16:44, 3 August 2016 (UTC)
✓ Done That was done some time ago but I forgot to update that topic. Dachary (talk) 12:15, 14 September 2016 (UTC)

instance of (P31) qualifier of source code repository (P1324)[edit]

@NMaia: you added mirror (Q654822) as a qualifier to a source code repository (P1324) at https://www.wikidata.org/wiki/Q20085696 which sounds like a great idea. Would you like to document that at https://www.wikidata.org/wiki/Wikidata:WikiProject_Informatics/FLOSS#source_code_repository ? If you don't have time, I can do it. Dachary (talk) 22:30, 2 September 2016 (UTC)

It's a good idea to say the website is a mirror but instance of (P31) is not a qualifier so p31 can't be used as a part of a claim that says something about the specific claim (and p31 say “this item is a specific example and a member of that class”; an url is not an item)
Maybe should we find a better way to write this ? — Metamorforme42 (talk) 23:12, 2 September 2016 (UTC)
I'm open to any ideas :) ~nmaia d 23:30, 2 September 2016 (UTC)

What about adding the qualifier archive URL (P1065) with the URL of the mirror ? Say git://foo.com/ has is mirrored at git://bar.com/ and git://frob.nitz, two archive URL (P1065) qualifiers are added to git://foo.com/, one with git://bar.com/ and the other with git://frob.nitz. There can be a source code repository (P1324) statement for git://bar.com/ but even if there is no such statement, the information is valuable and verifiable (to the extent that is is not a 404 URL at least). What do you think ? Dachary (talk) 09:07, 3 September 2016 (UTC)

Hmm, it's not exactly an archive though, is it? ~nmaia d 13:10, 3 September 2016 (UTC)
A mirror is only pertinent if it is online, like archives (an historic of mirrors including old mirrors is useless because a software repository can have a lot of different mirrors), but it is not exactly an archive (because up-to-date opposite to an archive who is a fixed representation of data at a precise time (we can't archive date (P2960) with a mirror)). Should we create a new qualifier (mirror) ? — Metamorforme42 (talk) 13:25, 3 September 2016 (UTC)
A new mirror qualifier would be better, sure. I agree with both of you that archive URL (P1065) is not ideal. I picked it because it already exists but it's definitely not a perfect fit. I would support a new mirror URL qualifier. Dachary (talk) 14:42, 3 September 2016 (UTC)
Me too. At any rate, it would be nice to get the wider community's take on the desirability of such a qualifier. ~nmaia d 13:38, 4 September 2016 (UTC)

✓ Done The discussion should move to the mirror URL property proposal created just now. Dachary (talk) 12:12, 14 September 2016 (UTC)

@NMaia:, @metamorforme42: it would be great if you could comment on the mirror URL property proposal. Dachary (talk) 09:10, 18 September 2016 (UTC)

@dachary: ✓ Done and I added french translation and subject item. — Metamorforme42 (talk) 14:08, 18 September 2016 (UTC)

propose a wikidata id to appdata.xml[edit]

Reach out to https://people.freedesktop.org/~hughsient/appdata/ to propose a new element with the wikidata id of the software : https://lists.freedesktop.org/archives/freedesktop/2016-September/000339.html Dachary (talk) 21:17, 7 September 2016 (UTC)

✓ Done The discussion continues in the mail threads above and will be concluded there. Dachary (talk) 12:14, 14 September 2016 (UTC)

removed feature (P756)[edit]

I've been using removed feature (P756) in a way which I'm not sure if it's correct, but it makes sense to me. In Devuan (Q18601928), I added removed feature (P756)systemd (Q286124). Similarly, in GNU Linux-libre (Q665683) I set removed feature (P756)Binary blob (Q763151). Any thoughts on this would be appreciated, thanks. ~nmaia d 13:42, 4 September 2016 (UTC)

Could you describe how removed feature (P756) should be used ? I think I get what you're after but it would help to have a detailed description. Thanks :-) Dachary (talk) 08:12, 7 September 2016 (UTC)
Well, that's what I was hoping you folks would help with. My reading of removed feature (P756) is something that is specifically removed from this edition/version/distribution. But I'd like to hear your understanding of it :) ~nmaia d 13:48, 7 September 2016 (UTC)
That's also my understanding. In the case of Devuan (Q18601928) you added removed feature (P756)systemd (Q286124) but it should also state from where it was removed. If used as a qualifier for software version identifier (P348) it would mean that it existed in previous versions and was removed in this software version. If used as a statement for a software that follows (P155) another software, it would mean something similar. IIRC Devuan (Q18601928) is a GNU/Linux distribution that has been created from scratch and had never had systemd to begin with. Maybe you meant to express the fact that it does not use systemd ? If that's the case, removed feature (P756) may not be the best choice. Dachary (talk) 20:09, 7 September 2016 (UTC)
Isn't Devuan based on Debian? ~nmaia d 21:35, 7 September 2016 (UTC)
It looks like Devuan is indeed based on Debian, my bad for not catching this. What about having removed feature (P756)systemd (Q286124) as a qualifier of the based on (P144) statement ? Dachary (talk) 07:00, 8 September 2016 (UTC)
Hmm that would be adequate, I think :) ~nmaia d 21:39, 8 September 2016 (UTC)
Cool :-) Would you like me to update the item and the documentation accordingly ? Or would like to do it yourself. Either way is fine with me :-) Dachary (talk) 11:53, 12 September 2016 (UTC)
It would be lovely if you could, I'm a bit overworked at the moment :P ~nmaia d 23:29, 13 September 2016 (UTC)
✓ Done Updated the software project page accordingly and you already updated the Devuan item. All good :-) Dachary (talk) 11:44, 14 September 2016 (UTC)

CVS :pserver: repositories are not URLs[edit]

There still are some CVS repositories around (for instance GNU Linear Programming Kit (Q838189)). Contrary to all other VCS, it is not a URL when presented as :pserver:anonymous@cvs.sv.gnu.org:/sources/glpk. We cannot use it as an object of source code repository (P1324) because a URL is required. Does someone have an idea on how to handle that ? Dachary (talk) 15:02, 14 September 2016 (UTC)

For Libtiff (Q1017110) I set the web page that has the :pserver: instructions as source code repository (P1324) and added both Hypertext Transfer Protocol (Q8777) and Concurrent Versions System (Q467252) as a workaround until we figure out something. Dachary (talk) 10:35, 15 September 2016 (UTC)
Apache Maven (Q139941) has already attacked the problem of encoding CVS connection methods (see http://cvsman.com/cvs-1.12.12/cvs_28.php) in to URIs (see https://maven.apache.org/scm/cvs.html). I am not sure we can leverage this solution or not (Concurrent Versions System (Q467252) does not actually use this syntax). Incidentally the Maven encoding solution does not produce a dereferencable Uniform Resource Locator (Q42253) but rather just an abstract Uniform Resource Identifier (Q61694). It should be noted that the Internet Assigned Numbers Authority (Q242540) URI scheme registry (see http://www.iana.org/assignments/uri-schemes) does not contain an entry for "scm". Also the registry mentions the "cvs" scheme (see http://www.iana.org/assignments/uri-schemes/prov/cvs) presenting another method which seems appropriate and seems to claim "Scheme creator: The CVS Team" (so perhaps this is the best encoding scheme). 50.53.1.33 04:16, 20 September 2016 (UTC)
Impressive research :-) What about we use http://www.iana.org/assignments/uri-schemes/prov/cvs as a URL and add a qualifier with the :pserver: string ? This way a human being can copy/paste the :pserver: string, we have a URL that is suitable for the source code repository (P1324) datatype and a bot can rebuild the :pserver: from the URL and verify they match. It's a little bit complicated but there only are a few dozens CVS repositories. Dachary (talk) 06:49, 20 September 2016 (UTC)
✓ Done The discussion continues with the proposal of the cvs scheme. Dachary (talk) 20:28, 26 September 2016 (UTC)

source code repository (P1324) protocol (P2700) qualifier[edit]

I am wondering about this. Most version control (Q189439) systems are not really instances of a communications protocol (Q132364) but they can and often do support multiple "wire" protocols for accessing a repository (Q3133368), so I asked myself: shouldn't this qualifier represent the protocol used to access a repository with a specified Uniform Resource Identifier (Q61694) or other such specifier (I notice discussions about Concurrent Versions System (Q467252) "wire" protocol specifiers not being URIs)? As examples, I cite:

And of course this is in addition to local file (e.g., file URI scheme (Q5448333)) which virtually all such systems must support. In general VCS are not required to support any network protocols (just the local file access though a distributed revision control system (Q1186723) that only supports local file access would probably be inane).

Perhaps the thinking is that the actual network "wire" protocol is already specified by means of the uniform resource identifier scheme (Q37071). That seems reasonable, however, I notice both of the above systems do not have instance of (P31) or other claims making them a communications protocol (Q132364) so creating qualifiers using them as the object of protocol (P2700) seems wrong. It seems to me media type (P1163) (e.g., "Content-Type: application/mercurial-0.1") or file format (P2701) would be a better property for qualifying the type of VCS repo specification claim. What do you think?

I ask this because I was reverted while making such changes to source code repository (P1324) protocol (P2700) qualifiers. Thank you for sharing. 50.53.1.33 15:00, 18 September 2016 (UTC)

  • You have a good point about the fact that URLs often support multiple protocols, regardless of the scheme they start with (for instance github.com URLs respond to HTTP as well as git). When relevant a source code repository (P1324) should be qualified with multiple protocols for that reason. The other points you make about protocols are valid but I'm not sure I understand the problem you're trying to describe. I'd be grateful if you could rephrase it to clarify. The URL of a source code repository (P1324) responds to a protocol (P2700), it cannot be defined by a file format (P2701) nor by a media type (P1163) which is a way to specify file formats and format contents transmitted on the Internet. (P.S. it would help with the dialog if you could create an account for yourself, otherwise the IP address you're using is likely to be different when you change location and it will be difficult to followup). Dachary (talk) 21:10, 18 September 2016 (UTC)
Well that is actually the point. A single URI only supports a single communications protocol (Q132364) as per its uniform resource identifier scheme (Q37071). However, multiple representations of a resource can be served at a single URI (e.g., via HTTP content negotiation (Q1128629)). Served resource representations can be content labelled (e.g., with media type (Q1667978), language tag (Q1059900), etc.). You mentioned GitHub (Q364), so I shall arbitrarily pick Lithium (Q6647924) as an example. The URL https://github.com/UnionOfRAD/lithium.git, as per its "https" scheme, can only be accessed via the HTTPS (Q44484) protocol (on port 443), however, it can, and does return different representations/contents based on the client request (e.g., HyperText Markup Language (Q8811), JavaScript Object Notation (Q2063), etc.). A URI with the scheme "http" is only accessible via HTTP (on port 80 unless otherwise specified) and one with the "git" scheme is only accessible via the insecure git protocol (on port 9418 unless otherwise specified; see https://git-scm.com/book/en/v2/Git-on-the-Server-The-Protocols). Mercurial (Q476543) is not really a protocol itself, however, it does support two protocols (in addition to local file access which can be represented via a URI with a file URI scheme (Q5448333)) specified by the following schemes: "http" (and secure "https") and "ssh" (see https://www.mercurial-scm.org/wiki/WireProtocol). Incidentally, though I usually prefer to contribute via IP, I do have an account (going back to 2004). 50.53.1.33 06:54, 19 September 2016 (UTC)
The protocol qualifier for source code repository (P1324) currently means : "what VCS can I use with this URL ?". This is not the precise definition of a protocol and you made it very clear, thanks for taking the time to explain in detail :-). We could either agree that this interpretation of the protocol qualifier, in this specific context, is good enough. Or we could propose a new qualifier that exactly convey the intended meaning. And after the qualifier has been created, a bot can change the protocol qualifier to use the new qualifier. Which way would you like to go ? Dachary (talk) 10:31, 19 September 2016 (UTC)
We need a property that corresponds to version control (Q189439) instead of communications protocol (Q132364) for qualifying such claims. I did a cursory investigation and it seems protocol (P2700) does have current multiple conflated meanings/usage (easily found from Wikidata:Property proposal/Archive/48#P2700 based on the original proposal for protocol (P2700)) so we do need a new property (we cannot just co-opt/update it to have the meaning we want) and will likely need to disambiguate the meanings/usages updating claims en masse via a bot as you mentioned. Incidentally, the new property should be able to be applied to any URL property (e.g., generic URL (P2699)) to specify its usage with a VCS (e.g., specifying links to branches or specific files and versions withing a repo). I recommend a property name along the lines of "version control system", "repository version control system" or "repo VCS type", etc. 50.53.1.33 02:15, 20 September 2016 (UTC)
Agreed. Could you propose such a property ? Once it exists I can add something to User:FLOSSbot to convert the existing protocol qualifiers of source code repository (P1324) to use it instead. Dachary (talk) 06:40, 20 September 2016 (UTC)

I'm browsing the properties looking for something to replace protocol (P2700). It should be more generic than media type (P1163) or file format (P2701). Here is what I found: including (P1012), uses (P2283) used by (P1535) or facet of (P1269) because the server reachable via this source code repository (P1324) URL has to include the VCS one way or the other to answer to the VCS client in a useful way; P794 (P794) because it is a valid VCS. While it makes sense when you think about it, there is very little chance an editor figures that out by herself/himself. I'll keep looking. Dachary (talk) 12:39, 21 September 2016 (UTC)

Well I notice there are bug tracking system (P1401) and package management system (P3033) properties. Surely, we can get a version control system property. That said, the target property needs to be a Wikidata qualifier (Q15720608) (which excludes uses (P2283), used by (P1535), facet of (P1269) unless we co-opt them a qualifiers; I believe very few properties are qualifiers and as properties on entities like items or other properties) and of type item (which excludes media type (P1163)). version type (P548) and file format (P2701) are interesting but I agree specifically in the wrong directions (and/or not general enough). including (P1012) also has this problem (too related to has part (P527)). P794 (P794) is general enough but perhaps too general (as you stated it would be hard to educate editors). 50.53.1.33 15:16, 24 September 2016 (UTC)
I agree :-) Would you like to propose the version control system property ? Dachary (talk) 22:17, 26 September 2016 (UTC)

Licenses or later[edit]

Should we create elements for GPL-1.0+, GPL-2.0+ and GPL-3.0+ with statements "subclass of GPL" and "has part GPL-3.0[ and GPL-2.0[ and GPL-3.0]]" ?

For example, Concurrent Versions System (Q467252) is actually under GPL-2.0+ (from the reference on the item) but writing GPL-2.0 is false because it is also GPL-3.0, and writing "GPL-2.0 and GPL-3.0" is only true for the moment (what if a GPL-4.0 come in some years? we lost an information (the +)).

Problem, this is not the only one item concerned: could we use a bot for checking in the references if the license is GPL-2.0+ instead of GPL-2.0 or, GPL-2.0 instead of "GPL-2.0 and GPL-3.0", or GPL-3.0+ instead of GPL-3.0, etc. (@Dachary:) ?

Problem number 2 (but it is not really our because we can't change software licenses, we just put the informations we have) : GPL-2.0+ is deprecated, GPL-3.0+ and GPL-1.0+ both.

Metamorforme42 (talk) 19:47, 24 September 2016 (UTC)

What do you mean by deprecated? The link does not help at all to understand what it means by that. ~nmaia d 20:31, 24 September 2016 (UTC)
It's only a syntax problem or a how to describe the license of a new project (see the bottom of this page), finally I don't think we are concerned. — Metamorforme42 (talk) 21:06, 24 September 2016 (UTC)

Idem for LGPL (LGPL-2.0+, LGPL-2.1+ and LGPL-3.0+) — Metamorforme42 (talk) 22:03, 24 September 2016 (UTC)

 Approved I think we should create elements for *GPL*+ as you suggest and https://spdx.org/licenses/ clearly explains why it is a useful way to describe a variation of the licensing terms of a software. Dachary (talk) 22:34, 24 September 2016 (UTC)
Ok, we have now GNU General Public License, version 1.0 or later (Q27016750), GNU General Public License, version 2.0 or later (Q27016752), GNU General Public License, version 3.0 or later (Q27016754), LGPL-2.0+ (Q27016756), LGPL-2.1+ (Q27016757), and LGPL-3.0+ (Q27016762).
I have now to check all softwares with gplv1only, gplv2only, gplv3only, lgplv2.0only, lgplv2.1only, or lgplv3only (✓ Done). So can a bot read references and determine witch item have not the correct license ? Or at least giving a list of the software witch have some of theses licenses with references. I can do it manually, but it will take hours. — Metamorforme42 (talk) 09:37, 25 September 2016 (UTC)
I could write something as part of User:FLOSSbot. How would you describe the algorithm in abstract terms ? Dachary (talk) 17:58, 25 September 2016 (UTC)
Thank you, I will probably describe it in the next few days (I have to sleep now…). — Metamorforme42 (talk) 20:22, 25 September 2016 (UTC)

✓ Done Since the implementation of scripts to better handle that is outside of the scope of this discussion. Dachary (talk) 16:26, 28 September 2016 (UTC)

Community edition and Enterprise edition[edit]

Have we already some elements like community edition or entreprise edition for use with applies to part (P518) on a license statement for example ? See Neo4j (Q1628290) witch for the moment use incorrectly review score (P444). — Metamorforme42 (talk) 20:17, 25 September 2016 (UTC)

It does seem like those would make good candidates for items as I am not aware of anything like that, however, I do note the similarity with software alpha version (Q2122918), software beta version (Q3295609), release candidate version (Q1072356) and Q12355314 (often used with version type (P548)). 50.53.1.33 20:53, 26 September 2016 (UTC)
Having different items (one for the Free Software version and one for the proprietary version) also makes sense to me (GitLab EE (Q25973915) and GitLab (Q16639197) are an example). It gets trickier when the developer gives the same name to both software despite significant technical differences (see Dual licensing above for instance). Using review score (P444) as is done in Neo4j (Q1628290) conveys the information, no doubt about that. But having separate items seems a lot simpler. Dachary (talk) 22:12, 26 September 2016 (UTC)
Ok, does exist a tool for duplicate an item easily ? Two items really look simpler (or maybe we could see this as two branches of the software ?) — Metamorforme42 (talk) 16:54, 27 September 2016 (UTC)
I'm not aware of a tool to duplicate items. Sound useful though :-) Dachary (talk) 20:20, 27 September 2016 (UTC)

For the record, I created Zarafa for home (Q27042442) today: it is the proprietary version of Zarafa (Q147680) and has a different name.Dachary (talk) 09:29, 28 September 2016 (UTC)

✓ Done I've handled a dozen such cases since the last comment and using the existing license / instance of and occasionally creating items when the software is different was enough IMHO. If someone is unsatisfied, please remove the done and let's continue the chat ;-) Dachary (talk) 20:26, 4 October 2016 (UTC)

License statistics[edit]

@Metamorforme42: The GPL-2.0+ etc. do not show at in the license statistics because they are not (even indirectly) a subclass of free software license (Q3943414) or open source license (Q1156659). I could add to the query that it should also include items that have a license that is an instance of a license which is a subclass of free software license (Q3943414) or open source license (Q1156659). Since this is getting a bit complicated, I would appreciate a confirmation from you that this is indeed the simplest way. Dachary (talk) 16:25, 28 September 2016 (UTC)

Imo, it is. And thanks to your modification, we can now see X11 license (Q18526202) in the stats. — Metamorforme42 (talk) 12:00, 29 September 2016 (UTC)
✓ Done Cool, thanks for the confirmation. Dachary (talk) 15:30, 29 September 2016 (UTC)

Heads up: Android ID proposal[edit]

Discussion on Wikidata:Property proposal/Android ID‎ is taking place, and your opinion on it is most welcome! ~nmaia d 18:27, 4 October 2016 (UTC)

Importing licenses[edit]

I'm implementing the following strategy to import licenses from various infoboxes, using User:FLOSSbot.

  • For each item that is an instance of (P31) Free Software that has no license (P275) claim
  • Parse each interlink to extract the license field from infoboxes
  • If the licenses of each interlink are exactly the same, add a license claim, otherwise display them for the editor to figure it out

Does that sound sensible ? Dachary (talk) 22:32, 8 October 2016 (UTC)

A first implementation was submitted for approval. Dachary (talk) 21:48, 12 October 2016 (UTC)

The problem with this implementation is that GPLv3 will be imported as GNU General Public License because it redirects to it. Handling of redirects should be done differently. Dachary (talk) 10:36, 15 October 2016 (UTC)
Looking at the wikipedia entries I think most of them link directly to GPL, even if the license infobox parameter states that it is GPLv3. When the version of the license is mentioned it is done in a way that I can't figure out how to parse. IMHO setting the particular version of the license should either be done manually or by harvesting information elsewhere. For instance via fossology, using the source code repository. Dachary (talk) 12:52, 17 October 2016 (UTC)

Property documentation[edit]

There is a lot of useful reference information about properties on this page, but the actual property documentation pages are decoupled from this -- see for instance Property talk:P1324 vs. Wikidata:WikiProject Informatics/FLOSS#source code repository. Wouldn't it be preferable if we moved this information to the property pages themselves, and link/transclude them here? Or alternatively, add them to a subpage of the property documentation, and transclude them in both places? --Waldir (talk) 22:39, 11 April 2017 (UTC)

Merge FLOSS with Software[edit]

Dachary
Metamorforme42
NMaia
Valerio Bozzolan
MichaelSchoenitzer (talk)
Jasc PL
LiberatorG
Pictogram voting comment.svg Notified participants of WikiProject Informatics/FLOSS MichaelSchoenitzer
dachary
Metamorforme42
Ash Crow
OdileB
John Samuel
Jasc PL
Daniel Mietchen


Pictogram voting comment.svg Notified participants of WikiProject Informatics/Software Could we merge page "FLOSS" into "Software", now that we have tabs in Software page? OdileB (talk) 10:27, 10 June 2017 (UTC)

Dachary
Metamorforme42
NMaia
Valerio Bozzolan
MichaelSchoenitzer (talk)
Jasc PL
LiberatorG
Pictogram voting comment.svg Notified participants of WikiProject Informatics/FLOSS MichaelSchoenitzer
dachary
Metamorforme42
Ash Crow
OdileB
John Samuel
Jasc PL
Daniel Mietchen


Pictogram voting comment.svg Notified participants of WikiProject Informatics/Software As nobody seems to object, I will probably try to do a merge FLOSS -> Software and copy the most informative texts and discussions into the still-non-existing Software page Discussion. I also wish to reorder the properties in alphabetic order. OdileB (talk) 15:03, 22 June 2017 (UTC)

Outdated software-versions by checking against Arch-Packages[edit]

Dachary
Metamorforme42
NMaia
Valerio Bozzolan
MichaelSchoenitzer (talk)
Jasc PL
LiberatorG
Pictogram voting comment.svg Notified participants of WikiProject Informatics/FLOSS

I wrote a little script that searches for outdated software-versions by comparing the newest version-number available on Wikidata to the version available in the Arch-Linux-Repos. You can find it's source-code here and it's results here. I plan an having a bot update it regularly – has anyone a bot that could do this? -- MichaelSchoenitzer (talk) 15:02, 18 October 2017 (UTC)

  • Good job! It may be a good idea to automate the adding of new versions of packages. ~nmaia d 23:45, 18 October 2017 (UTC)

Dachary
Metamorforme42
NMaia
Valerio Bozzolan
MichaelSchoenitzer (talk)
Jasc PL
LiberatorG
Pictogram voting comment.svg Notified participants of WikiProject Informatics/FLOSS

I've improved it a little. It now runs once a day on toolforge. It also tries to sort the list according to the "size" of the difference in version-number, so things where the first version number have changed are listed on top, etc. I think fully automatically adding the version-numbers from packages is not a good idea: this would lack the publication date (P577) and a source and there are a lot of false-posivitves (look at amule or wine for example). But what I would like to do, is to enhance the site of half-automatic editing formality, maybe a bit like the wikidata-games have them. I think, we should try to have a little team who looks at the list regularly and tries to keep at least all major-versionnumbers up-to-date. -- MichaelSchoenitzer (talk) 11:41, 28 December 2017 (UTC)

Ways to identify a Wikidata Item as Free/Open Source Software[edit]

Wikidata Free Software Identifiers Chart.svg

There are currently several ways that Items are identified as Free/Open Source Software:

Most of the time these are synonyms – but there are a handful of exceptions.

Currently there seems to be no plan or logic in what is used when. I generated a Venn diagramm to see how much they are used nowadays and how much they overlap. We should think about how we want to handle things in the future. I see several possibilities:

  1. Remove all the instance of (P31) statements and identify them only by there license.
  2. Use free software (Q341) wherever appropriate.
  3. Use free and open-source software (Q506883) when ever a software is free and open source software. Use free software (Q341) and open-source software (Q1130645) only in the rare cases when a software is only free (according to FSF) or only open source (according to OSI).

I personally strongly prefer Option 3, since it's the most correct one. Option 2 would be easiest to reach since its the most common of the three, but it's less correct than Option 3. Option 1 has no redundant information but has the problem that there are a lot of programs using there own (unnamed) license, so it would be hard/unrealistic to add them all to Wikidata as own items (at least for now). Also it makes accessing them harder. What's you opinion? Should we make an RFC? -- MichaelSchoenitzer (talk) 13:12, 5 January 2018 (UTC)

Tobias1984
Emw
Zuphilip
Danrok
Bene*
콩가루
TomT0m
DrSauron
Ruud Koot
Andreasburmeister
Ilya
Toto256
MichaelSchoenitzer
Metamorforme42
Pixeldomain
User:YULdigitalpreservation
Dipsode87
Pintoch
Daniel Mietchen
Jsamwrites
Tinker Bell
FabC
Jasc PL
putnik
Dhx1
Tris T7
Peb Aryan
lore.mazza004
Rc1959
Premeditated
Pictogram voting comment.svg Notified participants of WikiProject Informatics -- MichaelSchoenitzer (talk) 13:13, 5 January 2018 (UTC)

Dachary
Metamorforme42
NMaia
Valerio Bozzolan
MichaelSchoenitzer (talk)
Jasc PL
LiberatorG
Pictogram voting comment.svg Notified participants of WikiProject Informatics/FLOSS -- MichaelSchoenitzer (talk) 13:15, 5 January 2018 (UTC)

I totally agree on the fact that there are no plan or logic, as also for other wikidata arguments, so any effort to make a more effective approach is welcome,thanks. IMHO free and open-source software (Q506883) has a redundant definition: a free software must respect the four freedoms [1], and to respect the freedoms to study and to improve the program we need to have an "Open Source" code. free software (Q341) is comprehensive and should be used. Moreover, as you say, it is the most common term.
Talking about the licenses, Option 1 is my preferite way to proceed, since it would be easy for many contributors just to record the license type (GPL, PD, etc...). Then the license should be classified as instance of (P31) free software license (Q3943414).--FabC (talk) 14:10, 6 January 2018 (UTC)
Pictogram voting comment.svg Comment Agree with @FabC: that Option 1 is the best approach. To address the concerns raised by @MichaelSchoenitzer: about custom/unnamed licenses, those licenses should ideally be created as new items in Wikidata (with full work available at (P953) used to link to the license text if available), even if they're only ever linked to by one software item. License items can then be subclasses of a "free and open source software license" item. I'm sure there are multiple different interpretations of "free" and "open source" which this approach supports. Pixeldomain (talk) 00:02, 8 January 2018 (UTC)
Pictogram voting comment.svg Comment Trying to develop the concept to identify free software starting from their licenses (Option 1). This query lists the elements having a free software license, i.e. instances of subclasses of free software license (Q3943414):
# Free software and license
SELECT ?freesw ?freeswLabel ?freelicense ?freelicenseLabel WHERE {
  ?freelicense (wdt:P31/wdt:P279*) wd:Q3943414.      # Instances of free software licenses
  ?freesw wdt:P275 ?freelicense.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
ORDER BY ?freelicenseLabel
Try it!
Example:
NASA WorldWind (Q840415) license (P275) NASA Open Source Agreement (Q6952418)
NASA Open Source Agreement (Q6952418) instance of (P31) permissive free software licence (Q1437937)
permissive free software licence (Q1437937) subclass of (P279) free software license (Q3943414)
hence NASA WorldWind (Q840415) is a free software since it has a free software license.
The graph of the free software licenses can be shown here --FabC (talk) 18:40, 8 January 2018 (UTC)
Pictogram voting comment.svg Comment This only finds software under licenses listed as free software license (Q3943414) not open source license (Q1156659). See Wikidata:WikiProject_Informatics/FLOSS for a queries that finds both. -- MichaelSchoenitzer (talk) 01:28, 9 January 2018 (UTC)
Pictogram voting comment.svg Comment That's correct Michael, the query aimed to list only the items that can be considered pure "free software". --FabC (talk) 21:23, 9 January 2018 (UTC)