On 6/22/09 12:31 PM, "Gary Greene" ggreene@minervanetworks.com wrote:

On 6/20/09 5:43 AM, "Timothy Murphy" gayleard@eircom.net wrote:

...

I'm running a dovecot/IMAP server under CentOS-5.3 on my desktop, reading the mail with KMail on my laptop.

I have what seems a venerable and well-documented problem/bug; when I click on "Check Mail" I get an error message "Error while getting folder information" ... Actually, this doesn't appear to have the slightest effect on kmail, which works perfectly if one just presses Continue.

But I have worked out that the cause of the problem is that there exists a kind of ghost folder, "uidvalidity", which is listed among the folders on the kmail page but does not in fact seem to exist in my maildir on the server.

Deleting the folder under kmail has no permanent effect; it simply re-appears when I re-start kmail.

I assume the folder is listed in some way in the dovecot.index ; and my question really is: if I delete this index file will it be re-created automatically?

This bug/feature seems to have been present for years, which I suppose is par for KDE.

More like par for RH, since they focus on GNOME instead. (Can't wait to see when RHEL/CentOS when they base the KDE in it off the work of the Fedora KDE volunteers... Likely in 6.)

...

To make it as difficult as possible to diagnose, the error message on the kmail page is incomplete, even when expanded to the whole screen. In particular it does not specify (on the visible part) what folder is causing the problem. Nor is the error listed with other dovecot information in /var/log/messages .

Any suggestions/advice gratefully received.

Replying to my own message since I forgot to add... Can you look at the debugging output for Kmail in ~/.xsession-errors, since much of the KDE debugging channels get logged here.

Rex Dieter wrote:

...

...

But I have worked out that the cause of the problem is that there exists a kind of ghost folder, "uidvalidity", which is listed among the folders on the kmail page but does not in fact seem to exist in my maildir on the server.

Deleting the folder under kmail has no permanent effect; it simply re-appears when I re-start kmail.

I assume the folder is listed in some way in the dovecot.index ; and my question really is: if I delete this index file will it be re-created automatically?

This bug/feature seems to have been present for years, which I suppose is par for KDE.

Years? Bugs filed? If so, where?

At random, here is one from 2006: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=364953

When I google for "Error while getting folder information" I get reports going back for several years, without, as far as I can see, any clear resolution or even explanation.

Could very well simply be a bug in dovecot and/or incompatibility between dovecot/kmail.

Well, it seems to occur equally often under courier-imap .

I have solved the problem in my case, without understanding it.

KMail appears to show a folder, College=>inbox=>uidvalidity But I don't find any such folder anywhere in ~/Maildir/ . I've no idea if this phantom folder should be present, and if so what it means.

This folder had the setting, "Include this folder in mail checks". After unchecking this box (which I did not check in the first place) the problem goes away - I no longer get an error/warning when I Check Mail in KMail.

On Tue, Jun 23, 2009 at 10:53, Rex Dieterrdieter@math.unl.edu wrote:

Timothy Murphy wrote:

...

Rex Dieter wrote:

...

Years?  Bugs filed?  If so, where?

At random, here is one from 2006: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=364953

Timothy, did you read that web page?

An excerpt:

I figured out how this bug could be resolved. KMail has introduced a configuration option for imap folders that allows you to mount a lot of stuff as a KMail folder. The present default is to look for things to mount *everywhere* on the mail server. In my case KMail wrongly recognized some files/folders from my server as imap folders. When I tried to remove them I accidentally destroyed some data, so yeah this bug had been a pain. It took me a while before I figured out that this was supposed to be a "feature". It was under 'Namespaces' configuration within 'Modify Account' dialog box.

Did you try to solve your problem with the suggested approach above?

HTH, Filipe

On Jun 23, 2009, at 1:26 PM, Timothy Murphy gayleard@eircom.net wrote:

But having read the web-page, I still do not really understand. In my case there is a folder "uidvalidity" listed by kmail, but there is no file or directory with that name on the dovecot server.

Maybe it's an unrecognized imap response from the dovecot server that kmail takes as a folder name.

Take a look at your dovecot server to see if there are any advanced options enabled that might give a uidvalidity response.

-Ross

Gary Greene wrote:

Debian patches the hell out of KDE, and so do most other distros, this bug should not be called a "KDE" bug unless you've completely verified with vanilla source and have fully analyzed the code paths to assure that it isn't a distribution patch that screwed the pooch.

Huh? This bug has been seen on at least 2 different distributions (Debian and Fedora). There's no evidence at all of this being caused by a distribution patch, so until/unless evidence to the contrary is provided, this has to be assumed to be an upstream bug.

Kevin Kofler

Gary Greene wrote:

I again reiterate: have you tested this on a vanilla installation of KDE?

No. In fact I haven't personally tested this at all, the OP did.

Don't assume its an upstream bug until you have verified it by getting the upstream sources and testing it.

This demand makes no sense. Users don't have vanilla installations (they use the packages provided by their distribution) and are certainly not going to build KDE from source just to prove something which should be obvious to everybody except you (also considering the fact that it can be reproduced on more than one distribution). And packagers also have other things to do than building unpatched packages, also considering that it'd mean also building vanilla kdelibs etc. and also considering that sometimes the unpatched source doesn't even compile (e.g. because nobody tested it with the latest GCC which is used in Fedora) or is just broken in some other way. Plus, installing a properly-packaged vanilla KDE means replacing the distro one which is also something nobody wants to do.

As I said, most distributions patch KDE pretty heavily with vendor and community patches, some of which change behaviours (both defaults and code) from the expected ones from the upstream.

I know what distribution packagers do, I *am* a Fedora KDE packager. And we ask users to report their bugs upstream unless there is evidence that they're specific to Fedora. The vast majority of bugs which are reported to us are NOT specific to Fedora, they're upstream bugs. Also note that we don't even have that many patches, we try to stay as close to upstream as possible.

And as Fedora and Debian use DIFFERENT patches, trying to blame issues on patches when they can be reproduced in both distros is complete nonsense.

Please stop the bullshit, this is an upstream bug and needs to be reported and fixed upstream! In distro-specific bug trackers, it'll just bitrot, or you'll be asked to report it upstream if there's active triaging going on (which is what happens in Fedora, our triager or one of us packagers will ask the user to file the bug upstream).

Kevin Kofler (a Fedora KDE comaintainer)

On 6/23/09 8:39 PM, "Filipe Brandenburger" filbranden@gmail.com wrote:

Hi,

On Tue, Jun 23, 2009 at 20:41, Kevin Koflerkevin.kofler@chello.at wrote:

...

...

I again reiterate: have you tested this on a vanilla installation of KDE? Don't assume its an upstream bug until you have verified it by getting the upstream sources and testing it.

this is an upstream bug and needs to be reported and fixed upstream!

Although I agree that most bugs in KDE are upstream bugs and not distro-specific, in my recent experience what I have seen is that KDE is dismissing the bugs saying that the bug does not exist in KDE 4 and that users should upgrade, which is obviously not an option for users of CentOS 4 or 5 (or RHEL for that matter).

I support an environment with about 25 users of CentOS (4) desktops, most of them used KDE and KMail (for historical reasons), but recently when they start bumping at those bugs I don't even bother and recommend that they move to GNOME and Thunderbird instead, which are much better supported, both in CentOS/RHEL, and in their own projects.

Some might consider KDE superior, but as long as they will refuse to support KDE 3, they are dead to me...

Filipe _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

Not to be a stickler about this, but GNOME did exactly the same thing when they went from 1.4 to 2.0. This is another case of major version pains that have bit KDE since the release of 4.0 that other major projects have had in the past. This is not to say that it is a good excuse, but it does happen...

As I told Kevin off-list, this is has caused some members in the KDE development community a certain amount of grief as bugsquad has started taking up the policy a few months back that you described above. (I personally feel the ball was dropped, and it wan't a good decision to cut out KDE3 support so early. I can see maybe killing off support for it three years after the last release, but not less than one year as has occurred.)

On Wednesday 24 June 2009 01:41:07 Kevin Kofler wrote:

I know what distribution packagers do, I am a Fedora KDE packager. And we ask users to report their bugs upstream unless there is evidence that they're specific to Fedora. The vast majority of bugs which are reported to us are NOT specific to Fedora, they're upstream bugs. Also note that we don't even have that many patches, we try to stay as close to upstream as possible.

This is contrary to what I've been told.

I always recommend trying to ascertain whether other distros also see the problem, and to report it upstream if that seems to be so. If it is not possible to ascertain that I was told that it should in the first instance be reported to the distro, who will then pass it upstream if relevant.

Has there been a recent change of policy?

Anne

Rex Dieter wrote:

The moral of my story is to prevent issues languishing downstream indefinitely, to avoid the kind of comment "this bug has been around forever, why is it still a problem and not fixed?", which does no one any good, esp when upstreams aren't made aware of the issues (which, by all accounts so far, seems to be what happened here in this particular dovecot/kmail case).

Dear Rex,

You are going to find this difficult to believe, but I did not actually understand your reference to "upstream" in response to my query (I was the OP).

Your comments always try to be helpful, but please remember that some of us wallow in a state of almost complete ignorance. It is impossible to insult our intelligence, and the simplest instructions or advice should be couched in language suitable for a 5-year old.

Timothy Murphy wrote:

Rex Dieter wrote:

...

The moral of my story is to prevent issues languishing downstream indefinitely, to avoid the kind of comment "this bug has been around forever, why is it still a problem and not fixed?", which does no one any good, esp when upstreams aren't made aware of the issues (which, by all accounts so far, seems to be what happened here in this particular dovecot/kmail case).

You are going to find this difficult to believe, but I did not actually understand your reference to "upstream" in response to my query (I was the OP).

Thanks for the whack with the reality cluestick.

My apologies, hopefully reading the following will help describe fedora's ideals and definitions here in more detail, http://fedoraproject.org/wiki/Staying_close_to_upstream_projects

-- Rex

Anne, the quoting on this got screwed, I'm not the one with the mail issue, since I build my own KDE builds and generally pull patches from the PIM Enterprise branch (I was the konstruct maintainer, so building from source for me isn't scary). I was commenting on the issue that Timothy was having.

On 6/23/09 6:45 AM, "Anne Wilson" cannewilson@googlemail.com wrote:

On Tuesday 23 June 2009 14:12:09 Rex Dieter wrote:

...

Gary Greene wrote:

...

On 6/20/09 5:43 AM, "Timothy Murphy"

...

But I have worked out that the cause of the problem is that there exists a kind of ghost folder, "uidvalidity", which is listed among the folders on the kmail page but does not in fact seem to exist in my maildir on the server.

Deleting the folder under kmail has no permanent effect; it simply re-appears when I re-start kmail.

I assume the folder is listed in some way in the dovecot.index ; and my question really is: if I delete this index file will it be re-created automatically?

This bug/feature seems to have been present for years, which I suppose is par for KDE.

Years? Bugs filed? If so, where?

Could very well simply be a bug in dovecot and/or incompatibility between dovecot/kmail.

I've been using kmail+dovecot for around 3 years now and never seen this ghost folder, so presumably there is some other influence at work.

Index files do get recreated automatically in kmail - but I thought Gary said that uidvalidity is listed among kmail folders? Index files are not listed among folders.

Gary - what have you done in the way of trying to locate this folder? Have you looked anywhere other than maildir? 'uidvalidity' sounds to me as though it might belong to an authentication scheme?

Anne