You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
$ curl "http://localhost:8000/api/some-resource?PYNO=ZA2134&PYNO=ZA2323&PYNO=100004232&PYNO=ZA1213" (the real list of PYNO values is much longer, abbreviated example.)
I expected the following
To get the result of the URL. 😄
With the -g / --globoff parameter, it works as expected.
I know that the behavior we have now has been this way for years, but... I still consider it a bad default personally, since I happen to use URLs with "array parameters" from time to time. It's also something that is easily supported in web frameworks like PHP. Given the "principle of least surprise", I think it would make more sense to invert the flag so that --globon is opt-in, for the cases when you want to work with URL sequences and ranges.
But maybe I'm the oddball here? I don't know, if people really like it this way, we shouldn't change it. On any occurrence, this would be a candidate for 8.x rather than something that should be done right now because we shouldn't make changes like this on a minor version release to avoid confusing people.
I think we should rather let  remain as-is even when globbing is enabled as a work-around. I don't think we benefit much from downright rejecting the example input.
curl supports and works with RFC 3986 URLs, and in such URLs there cannot legally be any [ or ] letters present since they are reserved and should be URL encoded to get used. The trend is however for browsers to move away from RFC3986 and so more and more "new" things find their way into the URLs that originally weren't.
curl has had globbing enabled by default for such a long time that there are now likely to be a very large amount of existing command lines and scripts using it, that changing the default is likely to cause grief to a large amount of users.
I think we should rather let  remain as-is even when globbing is enabled as a work-around.
Makes sense! If you'd point me in the right direction in the code base, I might even be able to submit a pull request for this. (No promises, but it's good nonetheless to document "where" and "how" that fix should be rightfully made.)