Rationale for this kind of system: If properly tended — and this can easily by done by committed users— a setup like this is a great tool for developers to get a clear idea of what the userbase considers to be high priority issues. The rating system gives voice to many users that might not feel like placing a comment otherwise. Here on the forums there are pile-on threads too, but they get lost amidst other threads and often become dormant after a while without it being clear what happened to it.

Yes, trac tickets are created sometimes and devs reference these tickets sometimes in the original thread, but in the current setup there is often a great disconnect between the user feedback and what is done with it — witness the threads referenced above.

I'm skeptical, but would be happy to be proven wrong.

Before we throw yet-another-tech (that happens to be used by a few other projects that have larger development teams), we should answer these questions:

What are the top communication issues that need to be addressed and is this tool actually help to address them? Who will be the uservoice wrangler when there is not really a trac wrangler?

Currently, users have a single point of communication with zotero (these forums) to address bug reports, feature requests, and support requests. These "brainstorm" webapps only really address feature requests. And the single thing they are able to do best is to see what popular opinion says are the most important issues. Developer priority is (and both should and must be) different than user priorities. Can one not already accomplish a similar thing by glancing at popular forum threads? Is the problem that this softwar solves even a real need for Zotero?

Migugg seems to most want a system to keep track of the next steps of program development. I don't know whether I agree to the importance of this (it seems more important that the scarce resource of developer time is spent improving Zotero). But I also don't think brainstorm software actually addresses issue tracking.

Mendeley's system actually demonstrates some of this disconnect: the second most popular request, over a year old, is for them to release their code under an open source license. The status is still listed as 'under review,' and the Mendeley developers only responded to it in the past few months to say "no, we can't do this."

Mark - thanks for this.
I haven't fully made up my mind, but I think I tend towards noksagt and ajlyon. Gathering and organizing feature requests doesn't seem a priority to me. And Dan has suggested that developers often don't feel they can make realistic predictions on issues, so respsonses might not be particularly useful.

Like miggug (and somewhat contra noksagt) I would place more value on issue tracking - I guess the tracker could work for this in theory, but does not in practice.

Personally, I'd be happy with something very simple - like a "to do" list of the core developers for the larger issues by order of priority - it wouldn't even have to include ETAs, but just knowing the next couple of things they are planning to do - and updating that maybe once every 3 months - wouldn't seem like a big use of developer resources and would already help - so basically a little more detailed roadmap than the one we currently have and I for my part would be content.

But I'm also perfectly happy to be convinced that a uservoice type forum can be useful.

Thanks mark for this. I do understand that programmers time is scarce and that there are good reasons why some issues are not adressed or delayed. BUT: A lot of time is also wasted by having discussions in the forums on issues that are not adressed, and as far as I can judge, could easily be adressed. Probably even more time is wasted by explaining why some issues are not adressed. This also applies to users. Countless hours are spent with asking for features and nobody knows whether this is even heard. I often wonder whether I should add a "+1" to a feature request, or whether this is not simply wasted time because I have no idea whether anybody listens.
If there were a system that would indicate to me if my request is a) heard and b) shared by others and c) taken care of (or, for good reasons not) that would help a lot.
noksagt's example of mendeley's feature of going open source is unfair. It is very obvious why they do not adress it, and noksagt just play's the obvious "bad commercial mendeley" card here. Apart from this issue, as far as I can judge, mendeley's system works perfectly fine and makes sense.

Dan, on:

As for the above, it's not lack of awareness of user priorities that prevents things from getting done—it's lack of developer time, and our having other priorities for a variety of reasons (including, most notably, grant requirements) that have nothing to do with how many users vote for a feature.

What do you think about ajlyon's point that ...

... the more problematic disconnect, I think, is between zotero-dev and Trac, and between patch submission and patch landing.

...? Is there something that can be done to address this?

Dan, about Redmine.

I'm currently using Redmine for my (bill-paying) job. I love it. It's the perfect balance of features you want without bloat. Running about 9 months with zero problems. A few admin and UI quirks but nothing major. I set it up primarily for my own use but colleagues liked it so much they've defined their own projects and are using it - actually more for project planning and task management, not specifically software development.

Contact me offline (see my profile) if you have questions or would like to hear more (I can give you an account on my system if you like). I myself installed and configured it and the database back-end. I'd be happy to help you guys out if I can.

Thanks all for your comments. To be clear, I agree with noksagt that "Developer priority is (and both should and must be) different than user priorities." The problem lies in the followup question: "Can one not already accomplish a similar thing by glancing at popular forum threads?" Looking at how this forum is presently organized, the answer is a qualified no. There is no straightforward way to find out what popular forum threads are. Even if you find one, there is no way to find out what the developer's stance is and whether it has been noticed (except by knowing the user accounts of the devs and spotting rare links to Trac). There is no feeling that it means anything to pile on with a +1 on such a thread.

I do agree with noksagt and others, including Dan Stillman, that perhaps this is not the greatest hurdle in Zotero's development cycle. Yet it is a palpable problem that quite a number of users have called attention to over the years. If Dan says, "As for the above, it's not lack of awareness of user priorities that prevents things from getting done", we are ready to believe him. Dan is taking the developers' point of view; fair enough. But from the users' point of view, there is an opacity-of-mind problem that can be quite annoying — a disconnect between what the devs are aware of and how the users can get to know what the devs are aware of. An ideas forum (where the status of ideas can be simply and clearly marked as "planned", "under review", "declined", with a brief explanation) helps solve this problem and in the end saves lots of valuable time.

I appreciate fbennet's insightful comments on the value of an ideas forum.

"...don't think the value of such a forum derives from any sense that users somehow control the development process. People who participate in digg-like systems are communicating with one another, as much as with dev central. To that extent, it can release some of the pressure to be recognized, and that can save time for everyone in the long run."

Also, his point about the use of such a voting forum to developers of third party plugins is an important one:

It's easy to come up with armchair visions for such things, but the necessary attention (for development, or for writing the grant applications necessary for development) can be galvanized by evidence of widespread demand.

In short, I'm glad some of you agree with the possible usefulness of an ideas forum like this. We'll have to see whether it flies or not. I'm waiting a bit to see if there's more input.

I guess it's worth asking what, concretely, an example like this offers that a traditional issue tracker does not. I guess I'm tending to lean towards noksagt's point on further reflection that it's "not much." Yes, it can give a way to gauge relative user interest in particular features. But that's a fairly small part of the issue here. The more fundamental questions are about transparent priorities going forward, about how work is organized, about how and how quickly patches get reviewed and committed (which is another way of saying how control of the code base is organized), etc.

What do you think about ajlyon's point that ...

"... the more problematic disconnect, I think, is between zotero-dev and Trac, and between patch submission and patch landing."


...? Is there something that can be done to address this?

The majority of the patches that have been waiting to land are translator patches. I try to get to them as I can (and I try to get to the high-priority ones immediately), but this isn't my department. We're simply without a dedicated developer to deal with translators at the moment. But, as I've noted, we're working on a better system for managing this process. (Trac isn't ideal for this.) I don't know the ETA for the new system.

In the meantime, though, I'm happy to give translator commit access to a few more of the usual contributors. We still need to push them to clients ourselves, but that's a bit easier once they're already committed.

For other patches, I would say it's not really any different from how you get a patch to land in most projects: try to get other people to test it, be persistent, and be patient.

I think giving ajlyon commit access for translators would be an obvious step at this point - he's atm the main person fixing them.

Developers can fail to comment in a brainstorm voting platform just as much as they can fail to comment in a forum just as much as they can fail to fully use an issue tracker.

Sure. As I mentioned, though, it's not necessarily the devs that should comment. If there was a source of up-to-date information (like a well-tended Trac), users could enlighten painful pile-on threads too. Some of you say "so improve Trac and the problem's gone". Well in fact users have been trying to disseminate whatever sparse information there is on Trac in various threads — we can just look at the loads of duplicate threads and the uncertainty in even the most popular pile-on threads to see how well that has worked so far.

This is not just about Trac. The problem is that these forums are a particularly bad place, interface-wise, to solve the disconnect I've observed above, because (1) there are no clear ways for users to quickly find what the devs have said about some topic; and as an immediate consequence, (2) many users just start a new thread if they can't find this information soon enough, wasting valuable time of many people including devs. The resulting duplicate threads make problem 1 only worse, reinforcing problem 2.

My perception of this is obviously different than the developers' perspective. I don't think Zotero is doing as good a job in communicating to the userbase as it could. Although the devs are confident they got the main priorities covered, the opacity-of-mind problem remains and this leads to a lot of time wasted in the forums. My intent was to suggest an improved interface that would help take away some of these problems. If this is not a priority right now, so be it. I look forward to the improvements that a better alternative for Trac will bring. I remain skeptical that the forums will become a much more enlightening place through it though.

I really like the fact that it essentially forces you to search before you can put in a feature request, I've often wished the forum did that.

Faolan actually implemented that for these forums months ago—we just never turned it on. We can, easily.

The main thing that's not clear to me is how another system (which, no, I probably wouldn't participate in) wouldn't become just as messy as the forums. It might seem cleaner initially when there's simply less content, but that would change if it were actually used. We're pretty aggressive about merging and closing threads here.

And the inability to properly deal with the inevitable support requests seems like a pretty serious problem to me, since one of the main points of these forums is that the developers can step in if need be.

there are no clear ways for users to quickly find what the devs have said about some topic

Given that we wouldn't be saying anything at all on a separate system, I'm not sure how well it would help with that, unless other moderators were willing to do a whole lot of quoting or linking. But then you're mostly just fragmenting discussion into a forum where developers participate and one where they don't.

And as can be seen for the endless list of repeating requests, people DO NOT find existing discussions.

But people do the opposite—post in irrelevant threads—just as a frequently as they fail to find relevant threads, and frankly posting in irrelevant threads is far more disruptive. Enabling auto-search will help with duplicates but will probably increase the frequency of irrelevant posts (which is one reason we haven't yet done it). I don't think we should overestimate the extent to which these can both be solved.

Also it would help, if somewhere on the entrance to the forums a note says: "if you want to push the development of a feature, simply add an entry with "+1"".

I don't think this is really a practice we need to encourage. It's mostly just distracting. We don't develop based on vote counts—we develop based on what we think will improve the software and what we have time to implement. It's usually clear when there's demand for a feature because it leads to actual thoughtful discussion.

Further, but probably this is technically not possible, a summary of the discussion on top of a discussion would help. It is quite annoying to find a discussion and then to discover somewhere in contribution 22 that a it is fix or a fix is planned or it can't be fixed.

I think this would be great and go a long way towards addressing some of the concerns here. This could be custom built, but maybe there's a Vanilla plugin to highlight particular posts in threads or add summaries to threads...

The other thing we've considered adding is a feature that gives some indication of the status of people posting, since being able to quickly tell who's a developer or power-contributor would probably be helpful to newcomers.

Quoting noksagt: "information than we have about Zotero's duplicate detection (which is: there is code that can be enabled by a hidden preference to find dups already, there isn't currently heavy development on that feature, but there are plans to improve this)."

Sorry to warp the thread, but do I read you correctly that there is a hidden preference to enable (some alpha version of) duplicate detection? Could you please give me a link to more information on that?

First: Zotero is wonderful. Thank you !
I am fully committed, which also means fully dependent on the functioning of Zotero.
If it went away, I'd be dead... So please keep up the great work :) !

That said, from my personal user experience, the forums leave a lot to be desired.
It can take hours to read the multiple and highly repetitive tracks that usually exist for popular topics. The forums are far too fractured - do not know where to go. If there is some useful information at all, it is often buried somewhere in the middle of a thread. Way too time consuming, and this time is to be multiplied with the number of users reading... No wonder many users avoid the search and open yet another thread.
Something like a summary, and more guidance/pruning would be highly welcome.

Feature requests are awkward. A search will lead to several threads, which may be on topic or not, have to be read completely, may have been answered or not, and the status and prospects often remain unclear.
And it is completely non-obvious what requests are popular to what extend.

Final aspect: Of course the developer point of view has to be different than that of the user. But somehow I seem to read here that the developers will go in the direction that they think the greatest and/or that is best for satisfying grant requirements, largely ignoring user opinion. Huh ??? Zotero is great, because it is so useful - to the user, of course... - and wouldn't that also the ultimate goal of a grant ?
One point in case is the lack of duplicate detection. Four years old in the forums, a "show stopper" to several people who commented, and by far the most popular request on Mendeley. A very simple detection on import would be enough for many people, as has been commented often - and still there is nothing. Why ?