kubuntu patches and backtraces

[Scott Kitterman]

Harald Sitter <apachelogger at ubuntu.com> wrote:
>alohas,
>
>I just realized that currently when we have to introduce a kubuntu
>specific patch these are not immediately obvious in backtraces, which
>is bad.
>
>What does everyone think about introducing a policy that
>a) all patches of upstream code must isolate their logic into a
>separate function
>b) that function must be prefixed with kubuntu_
>
>Benefits are:
>1) if our stuff crashes upstream knows that our stuff crashes
>2) if a crash appears deeper in the stack they know that somewhere we
>made changes
>
>Disadvantage:
>1) requires more code
>2) if upstream changes code we *modify* it may be less obvious that
>our modification requires updates
>
>For example in kde-workspace we have
>kubuntu_avoid_zic_and_deep_copy_timezone_data.diff which currently
>looks like this:
>
>+        } else if (!QFile::symLinkTarget(tz).isEmpty()) {
>+          if (!QFile::copy(QFile::symLinkTarget(tz),
>"/etc/localtime")) {
>+            ret |= TimezoneError;
>+          }
>
>with the new policy this would be
>
>+        } else if (!QFile::symLinkTarget(tz).isEmpty()) {
>+          kubuntu_avoidZic(tz);
>+          }
>
>HS

But only for long term, not upstreamable changes, right?

Scott K