Blazegraph fork?

[berezovskyi]

Now that it seems Amazon has acquired Blazegraph (https://www.trademarkia.com/blazegraph-86498414.html etc), does it make sense for the OSS community to fork Blazegraph and restart the development (at least processing the PRs)?

[ianwdunlop]

Another goal of a forked/reborn blazegraph would be to migrate the sesame dependency to org.eclipse.rdf4j.

[ianwdunlop]

Anyone still interested in forking blazegraph and trying to keep it alive in the OSS community? Maybe someone already has. If so where is the project? Maybe a possible strategy is to fork and try to build from master and the latest unreleased branch (maybe 2.2.0). I've tried building but there are some test failures. Has anyone else tried building from source?

[beebs-systap]

@ianwdunlop @berezovskyi We've had some recent discussions on this and there seems to be community interest, which we're working to gather together. Feel free to update this interest for folks that would be interested.

@smalyshev @thompsonbry

[nyurik]

I propose we set up a project, e.g. bigblaze or blazebig seem to be non-trademarked, and add anyone interested from the community as members. Of course it would be great if the original authors of Blazegraph would participate in this development.

[ianwdunlop]

@nyurik That's a fair proposal. It would be good to hear more from @beebs-systap and folks about what their intentions are - whether they would like to keep the original blazegraph repo going or if it is time to start a new project based on the existing code.

[thompsonbry]

There have been some discussions about moving the project to wikidata / wikimedia and some strong expressions of interest for that from the their community. Adding Stas and Beebs..

Bryan

On Wed, Dec 5, 2018 at 01:55 Ian Dunlop notifications@github.com wrote:

@nyurik https://github.com/nyurik That's a fair proposal. It would be good to hear more from @beebs-systap https://github.com/beebs-systap and folks about what their intentions are - whether they would like to keep the original blazegraph repo going or if it is time to start a new project based on the existing code.

— You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub https://github.com/blazegraph/database/issues/86#issuecomment-444426488, or mute the thread https://github.com/notifications/unsubscribe-auth/ACdv4Om1bJrQQYFtPT0wwL5mEC5_CeoDks5u15ghgaJpZM4TSFRN .

[nyurik]

I think @smalyshev is doing an amazing job with various Blazegraph improvements, and certainly should be one of the key people driving this project forward, but I do not think Blazegraph fork should be under Wikimedia's umbrella, using less convenient or familiar tooling (gerrit/phabricator), or managed by WMF management. Rather, I think it should remain in GitHub e.g. a separate organization as a typical open source project, with a low barrier of entry, a wide community of contributing and reviewing developers, and WMF developers actively participating.

[thompsonbry]

I suspect that the project could stay where it is in terms of tooling. The question is more about the responsibility for producing releases. Brad has discussed some options for providing CI resources.

It would be great to get a sense of how many people would be interested in leaning in to help wikimedia support the open source blazegraph platform. That might help them make this transition happen.

Thanks, Bryan

On Wed, Dec 5, 2018 at 7:54 AM Yuri Astrakhan notifications@github.com wrote:

I think @smalyshev https://github.com/smalyshev is doing an amazing job with various Blazegraph improvements, and certainly should be one of the key people driving this project forward, but I do not think Blazegraph fork should be under Wikimedia's umbrella, using less convenient or familiar tooling (gerrit/phabricator), or managed by WMF management. Rather, I think it should remain in GitHub e.g. a separate organization as a typical open source project, with a low barrier of entry, a wide community of contributing and reviewing developers, and WMF developers actively participating.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/blazegraph/database/issues/86#issuecomment-444535546, or mute the thread https://github.com/notifications/unsubscribe-auth/ACdv4DEpGkS2h7KNyYWDf_IGmP6xjAPMks5u1-wWgaJpZM4TSFRN .

[turbolent]

In https://github.com/turbolent/embergraph I've took 2.1.5-RC and started renaming the code to Embergraph (to avoid inflicting on trademarks), cleaning up the code base, and updating it (Java 8).

I plan on mostly updating the SPARQL server (probably replacing Jetty and servlets with something else) and will try to update dependencies (e.g. migrate to RDF4J).

[thompsonbry]

Awesome for getting involved. However, you might consider just issuing CRs. Stas should have access to a CI setup and he and/or Brad can cut releases.

Bryan

On Sat, Jan 12, 2019 at 22:51 Bastian Müller notifications@github.com wrote:

In https://github.com/turbolent/embergraph I've took 2.1.5-RC and started renaming the code to Embergraph (to avoid inflicting on trademarks), cleaning up the code base, and updating it (Java 8).

I plan on mostly updating the SPARQL server (probably replacing Jetty and servlets with something else) and will try to update dependencies (e.g. migrate to RDF4J).

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/blazegraph/database/issues/86#issuecomment-453806902, or mute the thread https://github.com/notifications/unsubscribe-auth/ACdv4DUIYPRwxxyzwvX7u44rTqdYgcwyks5vCtdXgaJpZM4TSFRN .

[thompsonbry]

CR == pull request

On Sun, Jan 13, 2019 at 16:41 Bryan B. Thompson thompsonbry@gmail.com wrote:

Awesome for getting involved. However, you might consider just issuing CRs. Stas should have access to a CI setup and he and/or Brad can cut releases.

Bryan

On Sat, Jan 12, 2019 at 22:51 Bastian Müller notifications@github.com wrote:

In https://github.com/turbolent/embergraph I've took 2.1.5-RC and started renaming the code to Embergraph (to avoid inflicting on trademarks), cleaning up the code base, and updating it (Java 8).

I plan on mostly updating the SPARQL server (probably replacing Jetty and servlets with something else) and will try to update dependencies (e.g. migrate to RDF4J).

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/blazegraph/database/issues/86#issuecomment-453806902, or mute the thread https://github.com/notifications/unsubscribe-auth/ACdv4DUIYPRwxxyzwvX7u44rTqdYgcwyks5vCtdXgaJpZM4TSFRN .

[beebs-systap]

@turbolent Would you be amenable to setting up a CLA to push the changes back into the core Blazegraph?

[ianwdunlop]

2 quick comments. Why do we need a CLA to push changes (via pull request I assume) back into the core blazegraph repo? Is this a different repo from this one? Isn't all this covered by the current licence (GPL 2)? From recent comments it seems that you would like to restart blazegraph development using this repo rather than fork off to anywhere else. I like that idea.

[nyurik]

In https://github.com/turbolent/embergraph I've took 2.1.5-RC and started renaming the code to Embergraph (to avoid inflicting on trademarks), cleaning up the code base, and updating it (Java 8).

I plan on mostly updating the SPARQL server (probably replacing Jetty and servlets with something else) and will try to update dependencies (e.g. migrate to RDF4J).

I do like the Embergraph name, but there does seem to be another project by that name. Not sure if that's a problem.

I am not sure what you meant by renaming the code to Embergraph - IANAL, but it seems MariaDB who went through the similar process did not modified the actual code - as that would make the code incompatible with extensions/external modifications. There is a difference between classnames that use Blazegraph or Bigdata internally, and the branding of the product. I suspect we only need to modify the later.

[beebs-systap]

@ianwdunlop @nyurik Yes, you can definitely fork it under GPLv2 into a separate project. However, we would like to see if there's interest in getting a set of interested parties to collaborate on restarting development in this github. WMF/Wikidata would definitely have a role as a significant stakeholder, but is not the only one. The CLA is needed to make sure that the contributions can be accepted under GPLv2, but it can be a fairly lightweight process. Would you be amenable to setting up a quick chat to discuss in more details?

[nyurik]

@beebs-systap I obviously would prefer not to fork if possible, and sure, ping me at my firstandlastname at gmail :) I was under the impression that now that the trademark belongs to Amazon, no one outside of Amazon can use it, hence the need for a fork. Another (possibly minor or irrelevant?) concern is re-licensing - e.g. if a person contributes code to this repo under GPL2, can Amazon re-license that contribution under a proprietary license?

[ianwdunlop]

I agree with @nyurik that keeping it in the original repo is preferred over a fork. If a CLA is necessary to contribute under gpl2 then that seems ok. I imagine what contributors need assurance over is that any code they commit remains open source forever. I guess we would have to clarify any trademark issues since that could cause problems for some people. Ping me at gmail for a lightweight discussion. One thought that did occur to me is what branch development should be focussed on? master or the latest unreleased branch or another? What would be the priorities for a re-invigorated blazegraph? I think it would be getting dependencies up to the latest rdf4j etc rather than focussing on new features.

[beebs-systap]

@ianwdunlop @nyurik Sounds workable. Yes, everything would be open source under GPLv2 and the CLA is a fairly standard process (it has always been in-place for Bigdata/Blazegraph). I think there's two priorities; updating the 2.1.5 branch as needed and picking up the 2.2.0 branch for future work.

Let me setup a quick meeting/discussion offline with you and we can chat through some of the details.

[thompsonbry]

We had some discussions at ISWC (last fall) with wikidata and some others on priorities. Priorities should emerge out of the community, but clearly updating the dependencies would be high on the list based on previous discussions.

Bryan

On Wed, Jan 16, 2019 at 8:54 AM Brad Bebee notifications@github.com wrote:

@ianwdunlop https://github.com/ianwdunlop @nyurik https://github.com/nyurik Sounds workable. Yes, everything would be open source under GPLv2 and the CLA is a fairly standard process (it has always been in-place for Bigdata/Blazegraph). I think there's two priorities; updating the 2.1.5 branch as needed and picking up the 2.2.0 branch for future work.

Let me setup a quick meeting/discussion offline with you and we can chat through some of the details.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/blazegraph/database/issues/86#issuecomment-454854244, or mute the thread https://github.com/notifications/unsubscribe-auth/ACdv4Obz0zl7R-ADB7n-K5izp81BWLDPks5vD1kvgaJpZM4TSFRN .

[berezovskyi]

I am not sure if depending on Amazon's goodwill regarding the trademark is a good idea. Also, CLA likely means a copyright assignment process which would allow Amazon to use GPLed contributions in Neptune without contributing any patches back to Blazegraph under GPL, which might not be in the spirit of GPL.

But I understand that Wikimedia and a few others are major users of Blazegraph and we should hear their opinion. From a cursory look, seems like @turbolent and @ianwdunlop have the most active forks and the former is free from trademark use.

UPD: USPTO/WIPO have 0 hits on 'Embergraph'.

[thompsonbry]

The CLA has been in place since 2006. We always accepted contributions under a CLA in order to help protect everyone’s rights. Of course, there are plenty of opinions about open source, but this is how blazegraph contributions were managed from the project inception. It is nothing new with Amazon.

Bryan

On Sat, Feb 16, 2019 at 10:53 Andrew Berezovskyi notifications@github.com wrote:

I am not sure if depending on Amazon's goodwill regarding the trademark is a good idea. Also, CLA likely means a copyright assignment process which would allow Amazon to use GPLed contributions in Neptune without contributing any patches back to Blazegraph under GPL, which might not be in the spirit of GPL.

But I understand that Wikimedia and a few others are the major users of Blazegraph and we should hear their opinion. From a cursory look, seems like @turbolent https://github.com/turbolent and @ianwdunlop https://github.com/ianwdunlop have the most active forks and the former is free from trademark use.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/blazegraph/database/issues/86#issuecomment-464371855, or mute the thread https://github.com/notifications/unsubscribe-auth/ACdv4GdxUoXKMM4OHU7ak8pvx15gUEn7ks5vOFO3gaJpZM4TSFRN .

[akuckartz]

Is that CLA published anywhere?

[thompsonbry]

I don't see it up online off hand. It is referenced at [1,2]. Brad would be able to provide you with a current version.

Thanks, Bryan

[1] https://wiki.blazegraph.com/wiki/index.php/Contributors [2] https://www.blazegraph.com/develop/

On Sat, Feb 16, 2019 at 10:38 PM Andreas Kuckartz notifications@github.com wrote:

Is that CLA published anywhere?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/blazegraph/database/issues/86#issuecomment-464422402, or mute the thread https://github.com/notifications/unsubscribe-auth/ACdv4KwThXV7Jc7VuiulGI-TfuExk9u2ks5vOPj1gaJpZM4TSFRN .

[berezovskyi]

These links may be of use to see how it looked in the past:

https://web.archive.org/web/20140103220007/http://www.systap.com/BigdataIndividualCA.pdf

and

https://web.archive.org/web/20140103215942/http://www.systap.com/BigdataCorporateCA.pdf

[elft3r]

Hi everybody, I'm currently in the process of evaluating different graph databases and one of them is blazegraph. It looks really promising, and we are thinking about choosing it for our solution.

However, the uncertainty of the future of this project, lets us still hesitate whether we should adopt it or not. Are there any updates about the plans of this project?

Thanks for your help, Jochen

[beebs-systap]

@elft3r Wikidata is running it in production and there's interest in continuing to move it forward.

[elft3r]

@beebs-systap Thanks for your quick reply. Are there already plans laid out on how it will move forward?

[beebs-systap]

@elft3r There's ongoing discussion and a general outline, which is in process.

[berezovskyi]

I am ok closing the issue given that there are new releases.

[ben-j-herbertz]

Hi all, what is the current status? There was no new release for almost one year ...

[ben-j-herbertz]

Well, as there is nothing new, should the project considerd dead?

I don't see it up online off hand. It is referenced at [1,2]. Brad would be able to provide you with a current version. Thanks, Bryan [1] https://wiki.blazegraph.com/wiki/index.php/Contributors [2] https://www.blazegraph.com/develop/

And those links are not working anymore...

In https://github.com/turbolent/embergraph I've took 2.1.5-RC and started renaming the code to Embergraph (to avoid inflicting on trademarks), cleaning up the code base, and updating it (Java 8).

This fork is also not maintained anymore...

I would really like to help here.

[berezovskyi]

Seems like teaming up with Wikimedia is the best course of action, their fork is very much alive:

• https://gerrit.wikimedia.org/r/q/project:wikidata%252Fquery%252Fblazegraph
• https://phabricator.wikimedia.org/search/query/jVcJQvQHKgRG/#R

[ben-j-herbertz]

So what should be the plan here?

Is there any difference between the wmf fork and this one?

[berezovskyi]

TBH, I only met one person from Wikimedia DE / Wikidata at a conference a few years ago who was not directly involved with Wikidata infra. We need to find a point of contact at the Wikidata infra team first to see if they are ready to lead the fork development (at least to package and publish it). Of course, would have been better if new maintainers could simply get write access to this org instead of forking...

[ben-j-herbertz]

I see, who has write access?

AFAIK the wmf fork is not really maintained as well..

[berezovskyi]

@beebs-systap and @thompsonbry I would presume.

[ben-j-herbertz]

Ok, i saw the jira was last maintained by @beebs-systap and @igor-kim (Jan. 2020).

But i dont know if one should work on the Tickets as, they are not maintained as well..

[ben-j-herbertz]

FYI: I got a simple CI Pipeline working (build the jar + Docker Image)

[beebs-systap]

@josn0w @berezovskyi I'd prefer to avoid a fork too. If you want to contribute, let's get a CLA in place and do it from within the current project. I'll reach out separately to discuss.

We'd also be interested in any others that also want to contribute.

[ben-j-herbertz]

@beebs-systap I would like to contribute here, if it is possible. It would be great if you can update on the current status of the Project and the plan for the future.

[beebs-systap]

@josn0w Great. Send me an email beebs [at] amazon.com RE: the CLA. The status of the project and the plan for the future depends on the contributors. We're happy to continue to support.

[jeffreyschultz]

any update on this discussion?

[berezovskyi]

@jeffreyschultz I am using Jena/Fuseki and the built-in SWI-Prolog RDF store for most of my project work now. Not sure if @ben-j-herbertz got push/releng/issue triage rights to the repo. I guess housekeeping like adding CI and updating CVE-affected dependencies would be the first step. Ensuring https://w3c.github.io/rdf-tests/ are covered would be another, though fixing issues like https://github.com/blazegraph/database/issues/234 would require some non-trivial commitment.

Wikimedia is evaluating migration off Blazegraph: https://phabricator.wikimedia.org/T206560