drkonqi and outdated software

[Reilly Brogan]

On 11/19/25 3:14 AM, Ben Cooksley wrote:
> On Wed, Nov 19, 2025 at 9:40 PM Harald Sitter <sitter at kde.org> wrote:
>
>     Hey!
>
>     I currently have an MR open to disable crash reporting to KDE by
>     default and would like your input on the matter.
>
>     https://invent.kde.org/plasma/drkonqi/-/merge_requests/364
>
>     Essentially we have a problem with users sitting on outdated software
>     versions, reporting crashes, and then getting their crash closed
>     because their version is no longer getting releases from us. So the
>     idea is we have a simple knob to turn on/off reporting to KDE at
>     compile time and leave it to distributions to toggle this as
>     necessary.
>
>
> Wouldn't the better idea be to have a "support end date" compiled into 
> the software, which defaults to the KDE value, and which is disabled 
> if a distro bug report URL is specified?

Why not have this mapping maintained elsewhere, like a file on the KDE 
CDN? Could be something like a JSON file mapping software components to 
minimum supported release or something like that and be automatically 
generated by CI from a release management repo? You could have it 
grabbed asynchronously by drkonqi after startup and then cached and then 
not re-fetched until the cached version was > 24h old or something like 
that.

That would allow for component maintainers to declare which versions of 
components they want to receive reports from and if for example kwin 
released a 6.5.3.1 hotfix they could add an entry that no reports from 
6.5.3 and below should be sent in. They could also perhaps use other 
metadata to exclude reports like perhaps distribution "foo" packages 
things in an unsupported way and they just don't want any reports from 
them at all.

This has the added benefit that it would only require modifications to 
drkonqi to implement the logic as well as figure out a way to expose an 
opt-out to the user. It wouldn't require changes to any other KDE component.

And obviously it would have a compile-time flag to disable the behavior 
for stable distributions. There will probably be some grumbling from 
those ultra-privacy-focused users about it but those users probably 
weren't sending in crash reports anyway.

Reilly
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/distributions/attachments/20251120/cd92e5ce/attachment.htm>