Organize accounts as follows:

Two levels of classifiers: Authoritative classifier, and Suggestive classifier. The server does not differate between them in any other way than marking classifications made by suggestive classifiers as suggestions. This information is made available to the clients. The clients used by authoritative classifiers could have a mode to walk through suggestive classifications and review them, and clients used by normal users could have a mode to hide suggestive classifications (if they are to many and to messy).

Instead of assigning an url to an answer, assign an arbitrary number of urls to the question.

A caching server/proxy, to “fan out” network traffic.

Ability to send and email each time a new question is added.

Ability to remove tags from server.

Ability to remove responses from a question (possibly with the constraint of there being no votes on them).

Ability to remove questions from server (with some constraints).

Empty votes should not be included in q_info_num_votes.

Ability to querry for questions with updates that a participant has voted in.

Add a default tag “Unclassified” to each new question and make each question public when submitted. Support sever-specific default tags.

Add timestamps for subelements in questions (tags, answers, winning response, total votes) .

The tags should be presented within a hierarchy (category, sub-category, etc.)

[done] A double-click on a tag or question should open the right item (without the need to click on a button)

[done] use a single-click to open the right item

Tags should be sorted in alphabetic order (following client localization)

Instead of having the question attributes (N, V, U) in this order, I would prefer to have them in (V, U, N) or (N, U, V) order because this would reflect the “amount of change” in the question: V (nothing to do on this question), U (consult the question to see what changed), N (read the question, vote, etc.)

We may have to discuss the meaning of N and U for tags. In particular, when a question is U, the tage should be U and not U+V. Proposal: only 2 letters: V and U.

Concerning organization of tags, it would be convenient for the classifier to have a software that helps identifying which tags to assign to a question in a quick way, typically by having them organized in some hierarchical way. This software can (should) be completely independent for the client.

[done] The ‘id’ field of the tag list doesn’t seem so useful. Something like the number of questions that have the tag would be much more useful, in my opinion ;

Maybe the ‘N’ and ‘V’ check boxes could be replaced by the number of new questions and the number of questions for which we still have to vote (or for which we have already vote) ;

[done] It would be nice if the ‘N’, ‘V’ and ‘U’ fields had tooltips (http://developer.gnome.org/doc/API/2.0/gtk/GtkTooltips.html) so that it would be easier to know what they mean.

The ‘id’ column of the question list doesn’t seem so useful neither. It could be deleted so that it would leave more space for the question text. The functionality of having the question ID could be replaced by an entry in a contextual menu which could copy in the clipboard the absolute URL of the selected question (the demexp:/blabla/id URL).

[rejected] Currently, when one wants to start using demexp, it has to run the client three times. First to configure the root login, then to create its own account, and finally to login with this new account. It would be much nicer if the user could use a « Wizard », launched the first time the client is run, to create its own account. d.: for any real server, the creation of accounts should be made by a centralized authority so this is not an issue.

There should be a convenient way, when you connect to demexp, to display in the central field (questions) the question that are new and the question that have been updated, and only those questions (for the moment you have to browse through all tags that are N or U)

When no tags are selected in client, all questions should be available.

When a user creates a new question, he/she should be able to vote immediately on it, without waiting for classifiers to classify it (otherwise it may cause frustration!)

The field that displays responses to a question should be bigger by default and its size variable.

When installing demexp client on a friend’s computer, its quite painful to ask him/her to vote on all questions. It would be nice to “import” all the vote from a delegate (typically yourself). I’m not sure it’s possible technically.

When logging onto your account from a friends computer, there would be nice to have a way to get your own tags attribute. One way of doing this would be to store your tag attribute file in an encrypted way on the server, when you loggout from your usual connection place

It would be nice to display the number of question and tags in demexp, somewhere in the client window. Just for fun

[fixed]When run with no bases.dmxp file, the demexp server doesn’t initialize the participant base with a default root login. See http://lists.gnu.org/archive/html/demexp-dev/2005-09/msg00082.html

[fixed in 0.7.x]When moving from one version to an other (like 0.6.1 to 0.6.2 or like 0.6.2 to 0.6.3), the cache is cleared and all questions are marked as new (again!).

When the a configure is made without server target, the make test should not try to test the server.

When changing the public/private status of several questions in row in classification window, the last status change is not made on the server.

a scripting language for silently querrying the server. Even more than that : something like Javascript for Firefox or lisp for Emacs.