Version: (using KDE 4.4.1) OS: Linux Installed from: Fedora RPMs There have been many discussions about this topic, but this bug is intended to consolidate the issues in one place. The KAddressbook implemented in 4.4 is missing many features which users have come to depend on in version 4.3 and prior: Missing fields - bug 222678 Missing Apply button - bug 222677 Confusing search results - bug 222676 Missing columns - bug 222690 No checkboxes to select addressbooks Missing SMS link in contact pane Inability to search across multiple address books ...more... The developer has stated that this version was implemented as part of the migration to Akonadi. However, the rest of KMail is not fully integrated with Akonadi, so the new address book integration is of limited utility at this point. Most of the users of KDE are not developers, with many of them also being business users who expect some degree of feature stability. These users are dependent on their distribution to provide package updates. The distributions are, in turn, dependent on the KDE project to provide updates. There is general understanding in the community that minor version releases (i.e. 4.3 -> 4.4) will not drastically break existing functionality. For the address book, this social contract has been ignored, causing a huge number of issues downstream to both distributions and end users. Here is a thread which describes a number of issues in detail: https://bugs.launchpad.net/ubuntu/+source/kdepim/+bug/507990 KDE is a great desktop environment and has produced a lot of technical innovation. However, there is sometimes an pattern of focusing too much on core technology and not enough on user experience. CalDAV support (bug 133614) is a good example of this. The issue has been open since 2006, with nobody to prioritize its implementation as a key competitive feature. When a simple KResource implementation was proposed, there were suggestions to use the fledgling Akonadi framework because it would be "easier". The focus should be the user, not the technology. We were able to engage an external resource who created the KResource connector in under a couple of person-weeks. The benefit is that the software is usable today on a mature platform. With the user experience as a guide, I ask that the community to please consider one of the following proposals: 1) roll back the addressbook implementation to the 4.3 version, or 2) accelerate development so that the addressbook is feature complete in 4.4.3. Akonadi integration promises to bring many benefits to KDE users. With a properly staged approach, the transition can be made easier for everyone in the community.
I have to respectfully disagree that this is a wishlist item. Functionality that already existed was severely broken. Using that logic, every bug that causes a regression is a wishlist item. I request that this please be reviewed by some other members of the community before making an assessment.
this is not a crash either, i put it as "bug" but please do One but report Per bug this is easier for developers to follow it. for 2) accelerate development so that the addressbook is feature complete in 4.4.3. this won't be possible because there is no new functionnalities in a stable release new features will be in kde 4.5
I appreciate your consideration. I also do understand your concerns about new features being added only in KDE 4.5. In almost any other situation, I would completely agree with that opinion. However, I feel that this is an exceptional case, since a large number of features were removed from a product that was working very well for the large majority of users. As such, perhaps this should be considered to be a bug fix rather than a feature improvement, since the removal of features is a major regression. There are many distributions that updated from 4.3 to 4.4, but will not update to 4.5 given their release cycle. It would be good to get the fix into the 4.4 branch so that all of those users can benefit.
Complete list of bugs filed for the currently known direct regressions: Missing fields - bug 222678 Missing Apply button - bug 222677 Missing columns - bug 222690 No checkboxes for address book resources - bug 233093 Missing SMS link - bug 221915 Inability to search across multiple address books - bug 233095 Editing resources causes resource to disappear - bug 233097 I have put references to this bug in each of the above bugs so the discussion about versions and priorities can be conducted in a single place. Thanks in advance for your help.
i think this is up to distributions to pick up trunk commits when needed and have feedbacks. This is harmfull to commit those important commits in branches
(In reply to comment #4) > Complete list of bugs filed for the currently known direct regressions: add Bug 234080 - kaddressbook does not store contacts to vcf
*** This bug has been confirmed by popular vote. ***
The latest bug is that my addressbooks are suddenly showing up as blank. They are remote VCF files, which were working perfectly through the 4.3 release. Can we please have a solid commitment to either fix the bugs or revert the addressbook back to the original version (removing Akonadi altogether)?
It turned out to be straightforward to revert the 4.4.2 addressbook to the version found in 4.3.5. Here are the instructions to build Kontact 4.4.2 with the stable addressbook: 1) Unpack kdepim-4.4.2.tar.bz2 2) cd kdepim-4.4.2 3) mv kaddressbook kaddressbook-alpha 4) tar xvfz kaddressbook-4.3.5.tar.gz 5) Build as normal The addressbook tarball is attached. It would be much appreciated if the KDE team could revert to the stable addressbook and wait until the Akonadi version is stabilized and feature-complete before including it into the mainstream distribution.
Created attachment 43212 [details] Stable addressbook
> i think this is up to distributions to pick up trunk > commits when needed and have feedbacks. You suggest that feature-incomplete software replace their feature-complete counterparts and be pushed to end users to get feedback? What feedback do you want? I personally compared Kaddressbook 4.3 to 4.4 and filed bugs that listed all feature regression, some of which were major. Want other feedback do you need from _end_users_ at that point? > This is harmfull to commit those important commits in branches Are you suggesting that pushing feature-incomplete experimental versions of otherwise stable software to end users is not harmful? Please tell me, what is the purpose of KDE SC, if not to provide usable software for the end user? From the kde.org homepage: "...dedicated to creating a free and user-friendly computing experience..." Shipping feature-incomplete software as a replacement for working software is user-friendly? From the "What is KDE" page: "Our community has developed a wide variety of applications for communication, work, education and entertainment. " How do you expect people to use the software for communication, work, education and entertainment when experimental feature-incomplete versions replace stable versions of software? From the "What is KDE" page: "Our products are used by millions of home and office workers, and are being deployed in schools around the world" Do you not think that you have some responsibility to the very home and office workers and schools that your flaunt as your users? From the "What is KDE" page: "KDE produces a number of key applications such as ... Kontact, the comprehensive personal information management suite." Kontact is billed as a key KDE application. Thus key functionality is removed from KDE when shipping experimental versions of software to end users. From the "What is KDE" page: "It is our hope and continued ambition that the KDE team will bring open, reliable, stable and monopoly-free computing to the everyday user." Reliable? Stable? Shipping this Kaddresbook version was the exact opposite of reliable or stable.
I'd like to add to Dotan's point above. It is impractical for a distribution to mix and match the address book, since it is an integral part of the Kontact build. If it were a separate package, it might be more feasible. My instructions for reverting to the 4.3 address book involve manually substituting a directory in the source tree. Also, Akonadi still seems to be alpha quality. There are all sorts of corner cases that have not been tested. For example, abrupt loss of access to the home directory (i.e. via NFS mounts) can cause database corruption, putting Akonadi in a bad state. Furthermore, the Akonadi management console has unpredictable behavior, such as hanging for indeterminate periods of time. This is not the kind of software to which users want to trust their precious personal data. Akonadi has created more problems that it has solved and really needs to be stabilized before the entire Kontact suite is committed to it. Maybe KDE needs to take an odd/even approach to releases like many other open source projects. Even releases are feature-complete and stable while odd releases are advertised as development branches. That way, distributions can stay on stable releases, getting necessary bug fixes along the way. I would have stayed on 4.3 for our users, but unfortunately Fedora stopped with 4.3.2 and skipped to 4.4.0 (now at 4.4.2). There were bug fixes that users needed from 4.3.3+, but they were not available to us. Therefore, we had to move to 4.4.0+, which introduced this unfortunate situation. For now, I have reverted the address book locally, but this is a maintenance overhead that is too high for most users.
Probably "putting an incomplete version in the so called 'stable' release, was the only way to have some feedback from users...". This could be the reason of this choice. Anyway I really hope that this experience of providing an incomplete software (compared to the previous version and which give some headcache to a lot of users) wouldn't be repeated again for the kmail to akonadi migration. Breaking some features of the addess book can be forgiven, but breaking the email application, for example, could be much more dangerous.
I do have the same problem in my installation of KDE 4.4.4. I fully agree to Kumaran, that the situation ist inacceptabel! - Stopping to release unfinished and untested software is the best way to return to a situation of stability - Removing fields in the address book that is used by thousands of users without giving them a way to save there data, is inacceptable Thanks to Kumaran that he put all the information together in this one bug report. For sure this is far more than a single bug. The question of development and release of the addressbook should be part of the central coordination of KDE development and release management. Regards, Martin
Hej, except of the 'Missing SMS link' feature (which is already work in progress), all issues mentioned in the initial report have been fixed in KDE-PIM 4.5. Sorry for the inconveniences in the meantime! Ciao, Tobias
It's good to see so many features back, especially the ability to sort on last name again. I see that categories are also viewable in the individual record, but I can't see any way to either filter or sort on them in order to identify a subset of the address book. Is this work in progress?