Switch to ArdorForum

Thanks for the update Evildave! Good communication is what the community needs. Future is bright! 

Thanks for the update Evildave! Good communication is what the community needs. Future is bright! 

Actually I didn't see anything new in Dave's communication, the hours of discussion resulted in the same outcome as at the beginning of this - "Use 1.5.15 if you can't use 1.6.2"

Can I suggest to restore confidence that the development team publish a Terms of Reference for future API changes, delimiting how the API will be changed, the communication people can expect and a promise around maintaining compatibility for 1 or 2 major versions of the API.

This is not just about the 'time' that the NXT devs have to do things, this is also about the time, money and effort that other businesses put into using NXT, for some of these NXT is a major part of their service so this is major inconvenience, for some NXT is just another crypto and here they may just delete it which harms NXTs adoption.

I've seen lots of posts around GNU licensing etc protecting the rights of the NXT source, so some posts protecting the rights of the users of NXT would also be nice to see.

Does that happen after the fork 1.7?
API which will be used?

Quote from: yassin54 on November 07, 2015, 09:31:52 am

Does that happen after the fork 1.7?
API which will be used?

That is still under discussion. Dave said there would be an update later.

hoo okok Thanks @cassius 

edit: I forgot to thank you for update @EvilDave 

guys I can see a direction where it all could go soon
current devs won't like it.
please stop doing such things, NXT DOES NOT belong to you.

Hmm, from the licensing used by Nxt, it belong to them more than you think coinomat, I believe tho. Maybe im wrong, im not a lawyer.

But of course, imo, thing will be better manage for the future. Everyone learn from mistake. In some sense, Nxt devs acknowledge that they screw up this time.

I'm pretty sure they will come up with some deprecated code management because that is the common sense.

Right, after a truly massive amount of communication/discussion/argument between myself, jl777, Damelon, Jean-Luc, Riker(formerly Lyaffe) and Chanc3r....the core devs and Nxt Foundation have arrived at a proposal package to allow Nxt, SuperNET and the wider Nxt-based eco-system to move forward from here.

We will be publishing the Slack logs for the entire discussion, so that everyone can see exactly how this proposal was created.

At this moment, it appears that the issues caused by the API change are under control and documented, and that most projects have adjusted to the changes. Good work, guys and girls.
Jean-Luc would like to apologise for any inconvenience caused to the Nxt/SuperNET community by this issue,

I'd like to quickly explain how this happened, afaik:
The API change was documented in the changelogs for the 1.6 branch, but the full consequences of the change for the API chain were not. Some people figured this out on their own, but some didn't. As the 'old' API calls were still active in the 1.6e test versions, the 'bug' did not appear for some project operators until the production version 1.6.2 went live.

This is a clear failure of the testing/review process for new Nxt releases, and we need to make some changes. Everyone involved wants to make sure that this sort of issue does not happen again in the future, and that no Nxt-based project is ever broken by an update of Nxt MainNet.

SuperNET was probably the most affected by this issue, so I've spent some time to get feedback from jl777 on how he would like to move forward, and the core devs will be meeting most of his requests, but not all.

So here are the main points:

The API will not be rolled back in a 1.6.3 release to the old 1.5.15 standard. The issues with 1.6.2 have been researched, understood and fixes have been implemented by most projects. (ordinary desktop users are NOT affected, just to remind everyone). The 1.6 API changes are vital to allow Nxt to move to the 1.7 branch, and will have to be implemented in any case. But: J-L has assured me that these are the last API changes needed for 1.7, which will have a release date around the end of this year.  There will be one small API change in 1.7, to do with the AssetDelete function, but there will be no other API changes, Nxt will therefore have have a stable API for at least the next 3-6 months, probably longer.

In order to ensure that there is a clear line of communication between Nxt core devs and external projects, we have set up an old-fashioned mailing list to allow a direct means of reporting bugs/requesting support/giving feedback from/to the core devs.  This will be a massive improvement on the current forum-based communication, and will only be open to serious devs and project managers. If you are one of those, contact Damelon or VanBreuk for more details.
Chanc3r has volunteered to lead on giving SuperNET support to help with any further issues that they may have, and the core devs are committed to helping everyone out with a smooth transition to 1.6.2.

The review/testing process for major releases will be changed. Bugfix/zeroday/patch releases will not be affected, but anything with an API change or a major new feature will have a compulsory review period of 2-4 weeks after release as an experimental version. In that time, the devs on the mailing list will examine and test the code to ensure that it functions in their production environment. (I'd like to see more use of TestNet here, btw.)
Once the new release has been checked out, the external devs will be asked to sign off and confirm that the new code is good for their purposes. This process will be purely about code testing and approval, not about feature requests.

On the feature request/planning side: J-L and Riker will be opening a seperate discussion about features (and particularly fee structures) for the upcoming 1.7 branch in the next couple of days. 

The API will not be rolled back in a 1.6.3 release to the old 1.5.15 standard. The issues with 1.6.2 have been researched, understood [...] The 1.6 API changes are vital to allow Nxt to move to the 1.7 branch, and will have to be implemented in any case.

Sad to hear that...

Quote from: EvilDave on November 07, 2015, 08:58:24 pm

The API will not be rolled back in a 1.6.3 release to the old 1.5.15 standard. The issues with 1.6.2 have been researched, understood [...] The 1.6 API changes are vital to allow Nxt to move to the 1.7 branch, and will have to be implemented in any case.

Sad to hear that...

I can understand that. I've spent a lot of time over the last few days (  ) talking to devs and trying to figure out a solution that keeps everyone almost happy, and that allows Nxt and SuperNET to move forwards together.
Both Jean-Luc and jl777 have perfectly good reasons to behave in the way they have, I can see both viewpoints clearly and they are both right.
But they are in very different worlds: Jl777 needs stability to build on (as do all other external projects) and the Nxt core devs have to advance, develop and improve Nxt.

Finding a balance between development and stability is what we need to do now, and the best way to do that is to implement better change and release management for the Nxt core, and to improve communications between Nxt core devs and external project devs.

Sign up for the mailing list, and start getting involved in this process.

Here is the transcript of the Slack conversation on this issue. Read it carefully.....

https://drive.google.com/file/d/0B4OEA-peIYHXd3JOTUY4VndQLW8/view?usp=sharing

It's only 58 pages.

Was nice to read, made me a much better idea of what's going on, I suppose I also need more james thoughts in that transcript to see his viewpoints.
Anyways, not being a dev, I agree with the current changes "which I understand that can be solved modifying some lines in the source code" or simply keeping 1.5.15
We all blamed bitcoin for being so slow in development, and now we start to cry because things go too fast? Come on... we are still so little... with this projects budgets, how can't they pay simply one guy to review this things and apply this api changes to the code? Wasn't supernets aim to innovate and make all cryptos work together instead of fight?
Seems very reasonable this change with upcoming singleton assets in 1.7, no one think in raspberry pi nodes? Mine already suffers with database stuff
We can't expect continuous api and also new features, want bitcoin with no new features on years? or nxt much features with consequent changes? When all of us entered nxt, I supposed which was due to innovation, so don't cry now.
For what I read in the transcript, JL had to acomodate a lot of code in order to work with new jetty and lucene releases, so, supposing that what he says is true "I repeat I don't know enought about of this", makes sense the other project developers spend some time adjusting their code also.

And the most important: The change to 1.6 was not mandatory either, so any bussiness had much time to check 1.6.2 before update "even I read the changelogs..." I agree with JL, doesn't makes sense to upgrade so happily to a major version and expect to work without any changes "everything need previous testing in large scale projects". And also agree with the need of supernet lite client to hardcode some trusted nodes, not random public ones "but this is other matter"

In conclusion, this problems "lack of testing of new releases and lack of communication between involved parties" seems to get solved with the recently agreed. So I hope this doesn't happen again, in my opinion was a fail from both parties and there was much egos fight, that is a thing which I hate to see, specially in so smart people like the ones involved in this issue, they left much reason in one side and go full ego mode... Spend more time in the other party side and try to understand their moves, please, both have reasons and faults, I suppose the aim of this is to work together, so don't destroy all the work already done splitting projects and fucking up everything...

Last thing: Good changes in 1.7 for fees, hope they also get lower for normal tx or very short messages, the high fee for permanent data is very reasonable having prunable transactions and in 1.6, the chance to recover it from other nodes "waiting for full news about this "

@tosh
Im not seeing much the overall goal of your post, in the end, the devs agree to review their release managment practice.
Can you be more specific?
I'm just trying to understand really.