We have silent, intermittent failures as well, using canjs. I investigated pretty deeply into StealJS, and tried to get some output from debug mode, but couldn't find anything.
Our build process runs over multiple apps at once (so our output looks a bit different), and it occasionally exits silently when opening one of the apps. For us, it looks like:
We run the build on a shared dev server, and the intermittent failures seem to be highly correlated with whether the server is busy running other tasks, and whether someone has the .js files open over a network connection. When it fails I can usually get a subsequent success by closing down the editor and waiting for dev to finish any heavy tasks it's churning through.
I tried, unsuccessfully, to reproduce this with a scaled-down app -- I ended up deciding/assuming that this was specific to our system, since our dev server can be a bit finicky at times. I haven't looked into it since then.
Does this sound similar to the failures you're running into?
Our build process runs over multiple apps at once (so our output looks a bit different), and it occasionally exits silently when opening one of the apps. For us, it looks like:
- $ steal/js steal/buildjs nns/page/* -packageSteal
- opening nns/page/account-create/account-create.js
- adding dependencies
- + nns/page/account-create/account-create.js
- // etc: if it echoes one dependency, it correctly finds and echoes them all.
- // we have 11 pages total, and it generally only fails on the ones with the most dependencies.
- opening nns/page/domain-overview/domain-overview.js
- $
- // that silent failure is treated as "success" by make
We run the build on a shared dev server, and the intermittent failures seem to be highly correlated with whether the server is busy running other tasks, and whether someone has the .js files open over a network connection. When it fails I can usually get a subsequent success by closing down the editor and waiting for dev to finish any heavy tasks it's churning through.
I tried, unsuccessfully, to reproduce this with a scaled-down app -- I ended up deciding/assuming that this was specific to our system, since our dev server can be a bit finicky at times. I haven't looked into it since then.
Does this sound similar to the failures you're running into?