incr.comp.: Make DepNode's std::fmt::Debug implementation useful again. #42625
Conversation
|
@bors r+ |
|
📌 Commit 5b5499d has been approved by |
|
@bors r- |
| let (def_id_0, def_id_1) = *self; | ||
|
|
||
| let def_path_hash_0 = tcx.def_path_hash(def_id_0); | ||
| let def_path_hash_1 = tcx.def_path_hash(def_id_1); |
There was a problem hiding this comment.
it'd be good to add a comment clarifying that this is a micro-opt, and the generic impl also suffices
…mance optimization
|
@bors r+ |
|
📌 Commit e323652 has been approved by |
arielb1
added
the
S-waiting-on-bors
|
⌛ Testing commit e323652 with merge 9430800... |
|
💔 Test failed - status-travis |
|
@bors retry (looks like a time-out) |
incr.comp.: Make DepNode's std::fmt::Debug implementation useful again. With #42537 a regular `DepNode` only contains an opaque hash as its identifier. In most cases, this hash is actually a `DefPathHash` and we can reconstruct the `DefId` it came from via a table lookup --- and then use that to print something intelligible for debug outputs. For cases where we cannot reconstruct information from the DepNode's hash, this PR will cache a string representation of the `DepNode` in a side-table. This string is later used for debug outputs. r? @nikomatsakis
|
💔 Test failed - status-travis |
|
@bors retry
|
|
Hm so the timed out build compared to the previous successful build shows that stage0-rustc jumped from 1000s to 2600s and stage1-rustc jumped from 1300s to 3500s. This PR doesn't seem particularly at fault, I'm messaging Travis to hopefully clarify |
incr.comp.: Make DepNode's std::fmt::Debug implementation useful again. With #42537 a regular `DepNode` only contains an opaque hash as its identifier. In most cases, this hash is actually a `DefPathHash` and we can reconstruct the `DefId` it came from via a table lookup --- and then use that to print something intelligible for debug outputs. For cases where we cannot reconstruct information from the DepNode's hash, this PR will cache a string representation of the `DepNode` in a side-table. This string is later used for debug outputs. r? @nikomatsakis
|
☀️ Test successful - status-appveyor, status-travis |
With #42537 a regular
DepNodeonly contains an opaque hash as its identifier. In most cases, this hash is actually aDefPathHashand we can reconstruct theDefIdit came from via a table lookup --- and then use that to print something intelligible for debug outputs. For cases where we cannot reconstruct information from the DepNode's hash, this PR will cache a string representation of theDepNodein a side-table. This string is later used for debug outputs.r? @nikomatsakis