Hi Tim, For production I'm building a new Smalltalk image (and a Docker image as well) on each commit to master in my Gitlab repo. I was suggested to add Lint check to the build process in order to avoid being caught by something like this again.
But the halt that got me was a notification from the Glorp cache telling the object that it was removed from it. So it was a particular condition that doesn't happen very often in my development image, and because I'm used to put halts here and there, ended up being commited and pushed to master. Best regards! Esteban A. Maringolo On Wed, Aug 19, 2020 at 3:24 AM Tim Mackinnon <tim@testit.works> wrote: > > Hi Esteban - its a good question (and I'm intrigued what can be done) - but > for production aren't you automatically building a fresh image with a CI > system... its pretty easy to do these days and this would stop that. > Additionally you could add something to search for self halt, as a build step > and fail your build on a non-zero result as well. > > Tim > > On Tue, 18 Aug 2020, at 9:46 PM, Esteban Maringolo wrote: > > Hi, > > > > I was bit by a bug in production I couldn't identify and it ended up > > being a lost halt in the code that was hanging my whole image. So no > > bug at all, a feature :-) > > > > Is there a way to disable the #halt and breakpoints? > > This way I could add such an expression (if existent) to my server > > startup and avoid issues like this. > > > > Regards! > > > > Esteban A. Maringolo > > > > >
