On Sun, Nov 29, 2020 at 3:51 PM Paul Gevers <elb...@debian.org> wrote:
> Source: topydo > Version: 0.13-3 > Severity: serious > X-Debbugs-CC: debian...@lists.debian.org > User: debian...@lists.debian.org > Usertags: issue > > Dear David, > > I have no idea what's happening, but currently we have multiple workers > (on different architectures) off-line on the ci.debian.net > infrastructure. In the current (short) list of pending tests, > todo.txt-base sticks out as being triggered (by topydo) some time ago > and waiting on all architectures. armhf finished (tmpfail) while I was > writing this message so I noticed it already ran multiple times but the > test on armhf times out after 2:55 hours (previous tests were under 1 > minute) with an error in autopkgtest so is treated as tmpfail (and > retried). > > Can you help us investigate what's going on? What is topydo doing? Topydo is a package implementing a todo.txt client. Todo.txt-base is a helper program adding features to todo.txt clients. Both are new, and I am new to the alternatives environment. I started by having todo.txt-base serve as the application installed as the todo/todo.txt application. Packages using this model were accepted, but turned out to be unworkable as I attempted to expand the ecosystem. So there are new versions of topydo and todo.txt-base in sid which use the todo.txt provider (topydo) as the home for the "todo" and "todo.txt" applications. It calls todo.txt-base features as necessary for the added features. Autopkgtests pass in sid. All works well, until you test the new version of one package along with the old version of the other. In one scenario there is a recursion, where topydo ends up calling itself via the alternatives hook. I didn't worry too much about this when I submitted the new packages, because the issue was transient, and the common popcon is like 5. I failed to consider that ci would evaluate the packages independently, and would choke. I'm reworking the packages so that they can manage this upgrade without complaint. That process will take at least days. I'd be willing to consider more drastic means to resolve this, either by overriding the failures this one time, or by deleting the packages from the distribution and starting fresh. -- AE0D BF5A 92A5 ADE4 9481 BA6F 8A31 71EF 3661 50CE