Finish up and release

[EXreaction]

What mods have anything to do with the template system?

It wouldn't be too difficult to run the automated installation tools in the Customisation Database to test installing any or all mods in the database on any version of phpBB, including 3.1, to see just how much is broken.

[Erik Frèrejean]

As 3.1 is based upon 3.0, most MODs should be able to be either compatible with 3.1 or very easily updated to work with it. Remember the core doesn't change very much, it just gets enhanced and keeping backward compatibility is a major thing (from the top of my head only the new request class actually requires a complete new way of interacting with the globals. That said request_var is kept fully functional for backward compatibility).

If I'm not mistaking again, new template engine was merged into 3.1 already. That means 3.1 got more major changes in comparison with 3.0 core, and thus MODs for 3.0 won't be compatible with 3.1 in that part and will have to be rewrited.

The public API of the template engine hasn't changed and therefore it is fully backward compatible therefore only MODs that modify the template engine itself break.

In conjunction with new extensions/hooks system, 3.1 should receive absolutely different way of modification and extensibility.

[/quote]
Sure but 3.0 MODs will in most cases still work

[Oleg]

One cause for the lack of external contributors is IMO the lack of releases. I personally lost interest in contributing code for 3.1 because I feel that I can spend my time better than writing code for a release that might be released somewhere in the future.

I looked at the list of commits you authored (https://github.com/phpbb/phpbb3/commits ... kfrerejean) and I did not see any major features. Are you unhappy that there is no release incorporating one of those commits yet? Of the merged 3.1 RFCs (viewtopic.php?f=81&t=35709) the only one that was not implemented by a development team member was the enhanced page titles one, and that diff looks trivially applicable to 3.0 if anyone wants to use it (as a matter of fact the diffs were based on develop-olympus). If I am overlooking anything please correct me.

Sure you'll always have board owners that won't upgrade, I've not decided that as well though most likely I won't be upgrading any time soon after the release. For my boards 3.1 doesn't bring anything we need any time soon (or not already have with MODs) but that isn't a reason to prevent the new features that are in the repository to be released.

This seems like a direct contradiction to the previous quote. If 3.1 has no functionality that you want why should we rush releasing it? By the time anyone gets around to contributing anything to 3.1 perhaps we will get the hooks done. (Yes, we do have contributions to 3.1, and if you are one of those contributors and you want 3.1 to ship with your contribution but without hooks please express this opinion.)

The issue with making frequent releases is that updating takes a significant amount of time due to modifications. Every release there are users bringing this up on .com support forums, and someone already brought it up in this topic. The reason why we want hooks in 3.1 is they will make updating infinitely easier for our users, at least in theory, and it is with hooks that we will be able to make quicker releases post-3.1.

It wouldn't be too difficult to run the automated installation tools in the Customisation Database to test installing any or all mods in the database on any version of phpBB, including 3.1, to see just how much is broken.

This will be a very useful and helpful thing to do.

[Erik Frèrejean]

Of the merged 3.1 RFCs (viewtopic.php?f=81&t=35709) the only one that was not implemented by a development team member was the enhanced page titles one

Well there is your problem, why aren't more community members implementing them? There are plenty of good coders in this community, but yet pretty much none of them can be bothered to contribute.

Sure you'll always have board owners that won't upgrade, I've not decided that as well though most likely I won't be upgrading any time soon after the release. For my boards 3.1 doesn't bring anything we need any time soon (or not already have with MODs) but that isn't a reason to prevent the new features that are in the repository to be released.

This seems like a direct contradiction to the previous quote. If 3.1 has no functionality that you want why should we rush releasing it?

3.0.6 also didn't have any new features that I needed/wanted as an administrator, but that doesn't mean that 3.0.6 didn't ship usable features for the community at large. The same applies for 3.1, it contains various enhancements and new features that the community want (otherwise there wouldn't be an RFC for them), but for my specific board I don't need them.
There is a difference between rushing releases and making sure that the community receives regular updates which enhance their board. If thinking of making a release lacking a feature that is being talked about for the last 2+ years and at the current pace wont be finished anytime soon is rushing, than yah rush it.

By the time anyone gets around to contributing anything to 3.1 perhaps we will get the hooks done.

Well this is exactly the point that is worrying me. A feature release gets pushed further and further back in order for a feature that may or may not be finished can make the release if it accidentally gets finished.

The issue with making frequent releases is that updating takes a significant amount of time due to modifications. Every release there are users bringing this up on .com support forums, and someone already brought it up in this topic. The reason why we want hooks in 3.1 is they will make updating infinitely easier for our users, at least in theory, and it is with hooks that we will be able to make quicker releases post-3.1.

For that reason I suggest extending the current hook implementation so that it actually becomes useful for MOD authors, it is already possible to create nearly edit less MODs (only the template edits remain) but because of the limited hook locations you need to use some nasty tricks to get things working. Overall a more powerful hook system is something phpBB needs/deserves to have, but is it really a show stopper?

Overall I think that the development is keeping itself hostage, a sensible development strategy was thought off but also thrown out of the window pretty fast. Using a feature (that isn't even in active development) as cut off point simply doesn't make any sense, and in the end when 3.1 finally gets released I don't believe that it will have the buzz justifying the development time (assuming it will be done before the summer you'll be talking of 2.5+ years).
If you want a selling point these days (like it or not) you'll need to have a steady flow of new features releases, users don't care about the size of the update but v300.54.9.199873 is way better than v3.0.10.

[nickvergessen]

...users don't care about the size of the update...

At least on phpbb.de they do. "The new Update is so big and breaks a lot of MODs, I can't install...." thats what you get as reply when you ask them to update on a support issue.
But nevertheless, with shorter release circles this should be fixed.

[Erik Frèrejean]

...users don't care about the size of the update...

At least on phpbb.de they do. "The new Update is so big and breaks a lot of MODs, I can't install...." thats what you get as reply when you ask them to update on a support issue.
But nevertheless, with shorter release circles this should be fixed.

Sorry I may had to be clearer on that, I mend they don't care when the update is relatively small. As you point out, most actually want that because it is less likely to conflict.

[Oleg]

Well there is your problem, why aren't more community members implementing them? There are plenty of good coders in this community, but yet pretty much none of them can be bothered to contribute.

I don't know. Why don't you contribute hooks so that we can release 3.1?

3.0.6 also didn't have any new features that I needed/wanted as an administrator, but that doesn't mean that 3.0.6 didn't ship usable features for the community at large. The same applies for 3.1, it contains various enhancements and new features that the community want (otherwise there wouldn't be an RFC for them), but for my specific board I don't need them.

3.1 has major changes to implement major new features. These changes break stuff. We have bugs filed against 3.1 where the corresponding functionality works correctly in 3.0. Stabilizing a major version takes time. We want 3.1 to be a meaningful update that users will want to update to. We don't want to release major versions too frequently because then we would be working on stabilizing them instead of implementing new features.

Considering that we release minor versions every half a year or so, having several years between major versions should not be a shock.

Well this is exactly the point that is worrying me. A feature release gets pushed further and further back in order for a feature that may or may not be finished can make the release if it accidentally gets finished.

There are two points here. One is we are going through the list of features for 3.1, therefore with time it is supposed to be shrinking (provided developers work on roadmapped features and not on new things). Two is we are actually making progress toward hooks. I reviewed the extensions diff this week, which is one part of hooks (and it is one of the "new RFCs" that was posted after the RFC freeze). With time we are more likely to get the hooks done than not.

For that reason I suggest extending the current hook implementation so that it actually becomes useful for MOD authors, it is already possible to create nearly edit less MODs (only the template edits remain) but because of the limited hook locations you need to use some nasty tricks to get things working. Overall a more powerful hook system is something phpBB needs/deserves to have, but is it really a show stopper?

While the development team has other priorities we are not removing old hooks in 3.1. Therefore you can post an RFC for enhancing the old hooks, and if you implement it before 3.1 beta (which you believe won't happen any time soon, correct?) those changes will be included in 3.1.

Overall I think that the development is keeping itself hostage, a sensible development strategy was thought off but also thrown out of the window pretty fast.

Not at all. We would still like to follow that strategy, but right now the reality is for 3.1 we cannot.

Using a feature (that isn't even in active development) as cut off point simply doesn't make any sense,

It is the top item on my todo list, besides reviewing and merging patches, which is pretty much as close as you can be to "active development" in a volunteer project.

With that, we are discussing here whether time has come to cut even hooks, and I am not yet sure that this is the case.

and in the end when 3.1 finally gets released I don't believe that it will have the buzz justifying the development time (assuming it will be done before the summer you'll be talking of 2.5+ years).

The point of hooks (done right) is that the development team won't have to bring modifications into the core, or implement everything that the community wants. Modification authors will be able to provide whatever functionality users want in a way that is both easy to install for users and easy to develop and maintain for authors. I think this is more important than any single feature the development team might implement.

If you want a selling point these days (like it or not) you'll need to have a steady flow of new features releases, users don't care about the size of the update but v300.54.9.199873 is way better than v3.0.10.

We can release 3.0.314159 instead of 3.0.10, but what does this really achieve?

[Erik Frèrejean]

Well there is your problem, why aren't more community members implementing them? There are plenty of good coders in this community, but yet pretty much none of them can be bothered to contribute.

I don't know. Why don't you contribute hooks so that we can release 3.1?

As I stated before

I personally lost interest in contributing code for 3.1 because I feel that I can spend my time better than writing code for a release that might be released somewhere in the future.

so far that hasn't changed.

3.0.6 also didn't have any new features that I needed/wanted as an administrator, but that doesn't mean that 3.0.6 didn't ship usable features for the community at large. The same applies for 3.1, it contains various enhancements and new features that the community want (otherwise there wouldn't be an RFC for them), but for my specific board I don't need them.

3.1 has major changes to implement major new features. These changes break stuff. We have bugs filed against 3.1 where the corresponding functionality works correctly in 3.0. Stabilizing a major version takes time. We want 3.1 to be a meaningful update that users will want to update to. We don't want to release major versions too frequently because then we would be working on stabilizing them instead of implementing new features.

Obviously, and the couple of bugs you have now will be nothing compared to the stream you'll get when the community at large starts to use them. The flip side of that last statement is the bigger the release, the more issues arise the more time it takes to stabilise. Overall I don't think that the overall time to stabilise for one big release is much shorter than that of various smaller releases. Presumingly even longer because there will be more issues.

For that reason I suggest extending the current hook implementation so that it actually becomes useful for MOD authors, it is already possible to create nearly edit less MODs (only the template edits remain) but because of the limited hook locations you need to use some nasty tricks to get things working. Overall a more powerful hook system is something phpBB needs/deserves to have, but is it really a show stopper?

While the development team has other priorities we are not removing old hooks in 3.1.

So there will be two hook systems in place? How would that work, does a MOD author have to go in and figure out which hook to use in which location?

Therefore you can post an RFC for enhancing the old hooks, and if you implement it before 3.1 beta (which you believe won't happen any time soon, correct?) those changes will be included in 3.1.

But that still won't speed up things. Or do you see this as a sort of shootout whichever one is finished first will be the cut off? If not the release still is hold back by the hooks RFC.

Overall I think that the development is keeping itself hostage, a sensible development strategy was thought off but also thrown out of the window pretty fast.

Not at all. We would still like to follow that strategy, but right now the reality is for 3.1 we cannot.

So far I haven't seen a compelling argument why it cannot. Its more that it is decided to not use it. As per my first post why not implement it at this moment? Any RFC not accepted or in progress (I guess that the hook RFC will be the exception) are pushed back. That way resources are freed up and you actually have a point to work towards.

One is we are going through the list of features for 3.1, therefore with time it is supposed to be shrinking (provided developers work on roadmapped features and not on new things)

This so far hasn't happened (this list only grew), and without actually freezing the list there is no way to make sure that the list will actually shrink.

[wGEric]

I don't know. Why don't you contribute hooks so that we can release 3.1?

You're attitude is incredibly concerning. I hope the whole development team doesn't share this same opinion. You are saying I'm going to do what I want and if the community wants something they will have to do it. You are trying to pass the blame when the development team is the cause of the problem.

Have you heard of the Pareto principle? "For many events, roughly 80% of the effects come from 20% of the causes." In phpBB's case, 80% of the development is done from 20% of the users. The 20% is going to be the development team. You are never going to have it where everyone is contributing an equal amount so quit trying to make it that way.

If the whole development team has the same opinion as Oleg then nothing is going to change.

You have people who have spent thousands of hours contributing to phpBB telling you there are problems and even offering suggestions on ways to fix it but your only response is to tell us to do more. We aren't people that joined yesterday. We have been part of the community for years. We have been or are currently team members.

Where is the rest of the development team? Where is Nils? Why hasn't your leader (or other managers) joined any of these discussions? Are they all posting in your private forums? I was a little upset to read that there was an internal discussion going on. It is very hard to have a discussion and resole issues if those that are in a position to make changes are having their own discussions and not participating with everyone else.

[sajaki]

Where is the rest of the development team?

they seem to be busy committing. I'm sure that we'll get some feedback soon.

https://github.com/phpbb/phpbb3/compare ... 4be9c70827