ontheweblop.blogg.se - Npm config set loglevel http

NPM CONFIG SET LOGLEVEL HTTP FULL

If it's not suppose to be catering to CI use, then I would kindly suggest that CI users set the env var themselves to their liking (which is trivial in virtually every CI system, as you mentioned). Question: Is this base image supposed to be catering to CIs at the expense of non CI use? Overriding npm's default is jarring and violates principle of least surprise. Of course, vocal support is usually in favor of change (the status quo rarely triggers noise).

the number of people who have commented on the original issue ( Excessive npm log output in 4.x #57) and this one seem overwhelmingly in favor of using npm's default.

virtually none of which actually prefer the docker image default) (so either preferring npm's default, or preferring it even quieter than npm's default.

actual use at my client, which shows hundreds of CI configurations, across hundreds of apps and libraries (and different teams with different opinions), ~90% of which force loglevel back to warn, the other ~10% of which force it to silent.

npmrc, instead of forcing users to override using env var or cli flag. Using this would allow users to force it back to npm's default using either user or project. $PREFIX/etc/npmrc is specifically for distributions to customize defaults to their liking.

NPM CONFIG SET LOGLEVEL HTTP FULL

it's possible to get the full debug log during a failure without being overly verbose to STDOUTĪt the very least, this setting should not be set via environment variable, but rather via the builtin config file.the vast majority of people are now forced to manually undo a setting for the tiny percentage that want it to be different than npm's defaultĪnd most importantly, contrary to the documented rationale for this setting: thus the only way to get back to npm's default is to set our own env var npmrc or npm config set because env vars take precedence.

indeed, failures themselves are the exception, so info is necessary in a fraction of a fraction of cases the scenarios where info is actually necessary to debug a failure are vanishingly slim. warn already prints (typically) enough info during a failure case.

conflicting with npm's own default setting is jarring and violates principle of least surprise.

Now that it's been over a year since this was last discussed, I would like to reopen the discussion to consider returning to npm's default. There are a long list of arguments why docker shouldn't be changing npm's default and a lot of vocal support for removing the verbose setting.