fix: no longer warns when no name was provided for global module #2359
Conversation
alan-agius4
force-pushed
the
feature_fix_global
branch
from
98ff468
to
d4267ef
Compare
alan-agius4
force-pushed
the
feature_fix_global
branch
from
d4267ef
to
ce856cf
Compare
alan-agius4
force-pushed
the
feature_fix_global
branch
from
ce856cf
to
2548760
Compare
|
|
||
| if (globalName) { | ||
| return globalName; | ||
| } |
There was a problem hiding this comment.
I think it's ok to not warn when the user has not specified any configuration, as in many cases this is the required behaviour that the package name be the request package name (eg this is the default on npm that a request package name and output package name can match).
The issue maybe arises in environments in which you are defining a global function or map that is trying to be authoritative and then has these invisible misses.
Instead, perhaps let's only warn for the case where globals is provided but has misses, by adding that as another case above?
There was a problem hiding this comment.
I am not 100% sure about this, as for UMD, I noticed that a lot of guessed name would be incorrect. If for instance having a . in the name. Or when the umd bundle name is totally different, and without a warning, it's hard to rollup users to know that something might be wrong.
However I can do what you suggested.
There was a problem hiding this comment.
@guybedford pushed an update.
alan-agius4
force-pushed
the
feature_fix_global
branch
from
2707448
to
69c9da2
Compare
src/Chunk.ts
Outdated
| return globalName; | ||
| } | ||
|
|
||
| if (hasExports || (hasExports ** globals && !globalName)) { |
There was a problem hiding this comment.
yeah looks like the IDE didn't save :P, re-pushed. Thanks for your patience 👍
alan-agius4
force-pushed
the
feature_fix_global
branch
from
69c9da2
to
10cd889
Compare
src/Chunk.ts
Outdated
| return globalName; | ||
| } | ||
|
|
||
| if (hasExports || (hasExports && globals && !globalName)) { |
There was a problem hiding this comment.
if (hasExports || globals && !globalName) would be simpler here I think.
|
I've been thinking about this some more and maybe you are right we should always warn actually... |
alan-agius4
force-pushed
the
feature_fix_global
branch
from
10cd889
to
2548760
Compare
|
hahah, removed the last commit :-)
…On Mon, 30 Jul 2018 at 14:13, Guy Bedford ***@***.***> wrote:
I've been thinking about this some more and maybe you are right we should
always warn actually...
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#2359 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AQv-WkwKWGdbBjdQnrha1iJ0YMqxPsogks5uLvhjgaJpZM4VlBf1>
.
|
|
LGTM |
When the dictionary lookup or function look up return undefined or null, a warning should be emitted and the module name should be guessed.
If the user, doesn't receive a warning that the global name was not found, they won't be able to fix a possible issue. If the global name is incorrect or later set to null.
Closes: #2358
Some more context
ng-packagr/ng-packagr#1026 (comment)