Commit 89ead6f
committed
fix(resolver): Treat unset MSRV as compatible
We last tweaked this logic in #13066.
However, we noticed this was inconsistent with `cargo add` in
automatically selecting version requirements.
It looks like this is a revert of #13066, taking us back to the behavior
in #12950.
In #12950 there was a concern about the proliferation of no-MSRV and
whether we should de-prioritize those to make the chance of success more
likely.
There are no right answes here, only which wrong answer is ok enough.
- Do we treat lack of rust version as `rust-version = "*"` as some
people expect or do we try to be smart?
- If a user adds or removes `rust-version`, how should that affect the
priority?
One piece of new information is that the RFC for this has us trying to
fill the no-MSRV gap with
`rust-version = some-value-representing-the-current-toolchain>`.1 parent e31c27d commit 89ead6f
1 file changed
+15
-32
lines changed| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
86 | 86 | | |
87 | 87 | | |
88 | 88 | | |
89 | | - | |
90 | | - | |
91 | | - | |
92 | | - | |
93 | | - | |
94 | | - | |
95 | | - | |
96 | | - | |
97 | | - | |
98 | | - | |
99 | | - | |
100 | | - | |
101 | | - | |
102 | | - | |
103 | | - | |
104 | | - | |
105 | | - | |
106 | | - | |
107 | | - | |
108 | | - | |
109 | | - | |
110 | | - | |
111 | | - | |
112 | | - | |
113 | | - | |
114 | | - | |
115 | | - | |
116 | | - | |
117 | | - | |
118 | | - | |
| 89 | + | |
| 90 | + | |
| 91 | + | |
| 92 | + | |
| 93 | + | |
| 94 | + | |
| 95 | + | |
| 96 | + | |
| 97 | + | |
| 98 | + | |
| 99 | + | |
| 100 | + | |
| 101 | + | |
119 | 102 | | |
120 | 103 | | |
121 | 104 | | |
| |||
271 | 254 | | |
272 | 255 | | |
273 | 256 | | |
274 | | - | |
| 257 | + | |
275 | 258 | | |
276 | 259 | | |
277 | 260 | | |
278 | 261 | | |
279 | 262 | | |
280 | 263 | | |
281 | 264 | | |
282 | | - | |
| 265 | + | |
283 | 266 | | |
284 | 267 | | |
285 | 268 | | |
| |||
0 commit comments