[lldb/Formatters] Remove space from vector type string summaries (NFCI)

[mib]

Summary

This patch changes the string summaries for vector types by removing the
space between the type and the bracket, conforming to 277623f4d5a6.

This should also fix TestCompactVectors failure.

Signed-off-by: Med Ismail Bennani <[e-mail]>

[teemperor]

Please also remove the FIXME, LGTM!

[mib]

Remove FIXME

[dblaikie]

Sorry I missed this - are these tested anywhere/should I have been able to discover if these needed to be changed before I made the change?

[teemperor]

In D112340#3081532, @dblaikie wrote:

Sorry I missed this - are these tested anywhere/should I have been able to discover if these needed to be changed before I made the change?

TestCompactVectors tests this but its unfortunately marked as Darwin-only (as it includes the Accelerate framework). Providing a platform-neutral test that just typedef's the same things as Accelerate (in addition to the test including the real framework) seems like a good idea.

[dblaikie]

In D112340#3081540, @teemperor wrote:

In D112340#3081532, @dblaikie wrote:

Sorry I missed this - are these tested anywhere/should I have been able to discover if these needed to be changed before I made the change?

TestCompactVectors tests this but its unfortunately marked as Darwin-only (as it includes the Accelerate framework). Providing a platform-neutral test that just typedef's the same things as Accelerate (in addition to the test including the real framework) seems like a good idea.

Good to know - will keep it in mind, but don't think I'm up for writing that test right now myself (not especially familiar with lldb test infrastructure, etc).

Was this reported by a buildbot, do you think/know of? Probably would've been happy to debug-via-buildbot and fix it that way, but don't recall seeing a fail-mail about this (but as always, there's a fair bit of buildbot noise and sometimes it's hard to find the cause/issue in a fail-mail such that I end up ignoring them)

[teemperor]

Not actually sure when/how Green Dragon is mailing external people,
but usually someone just watches the build bot and emails/comments on
commits that are breaking the buildbot. Green Dragon is more of a
monte carlo simulation than a real build bot so I wouldn't be
surprised if it doesn't send automatic failure emails to non-Apple
people.

Also that 'let's write a test' was more a note for myself to write
that test (which I hopefully get around today).

Am Mo., 25. Okt. 2021 um 19:10 Uhr schrieb David Blaikie via
Phabricator <[e-mail]>:

dblaikie added a comment.

In D112340#3081540 https://reviews.llvm.org/D112340#3081540, @teemperor wrote:

In D112340#3081532 https://reviews.llvm.org/D112340#3081532, @dblaikie wrote:

Sorry I missed this - are these tested anywhere/should I have been able to discover if these needed to be changed before I made the change?

TestCompactVectors tests this but its unfortunately marked as Darwin-only (as it includes the Accelerate framework). Providing a platform-neutral test that just typedef's the same things as Accelerate (in addition to the test including the real framework) seems like a good idea.

Good to know - will keep it in mind, but don't think I'm up for writing that test right now myself (not especially familiar with lldb test infrastructure, etc).

Was this reported by a buildbot, do you think/know of? Probably would've been happy to debug-via-buildbot and fix it that way, but don't recall seeing a fail-mail about this (but as always, there's a fair bit of buildbot noise and sometimes it's hard to find the cause/issue in a fail-mail such that I end up ignoring them)

Repository:

rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION

https://reviews.llvm.org/D112340/new/

https://reviews.llvm.org/D112340

[dblaikie]

In D112340#3084953, @teemperor wrote:

Not actually sure when/how Green Dragon is mailing external people,
but usually someone just watches the build bot and emails/comments on
commits that are breaking the buildbot. Green Dragon is more of a
monte carlo simulation than a real build bot so I wouldn't be
surprised if it doesn't send automatic failure emails to non-Apple
people.

Also that 'let's write a test' was more a note for myself to write
that test (which I hopefully get around today).

Cool - thanks for all the details!

lldb/source/DataFormatters/FormatManager.cpp

vector_flags.SetCascades(true)
       .SetSkipReferences(false)
       .SetDontShowChildren(true)
       .SetDontShowValue(false)
       .SetShowMembersOneLiner(true)
       .SetHideItemNames(true);
 
   AddStringSummary(vectors_category_sp, "${var.uint128}",
                    ConstString("builtin_type_vec128"), vector_flags);
-
-  // FIXME: These probably need the space removed, or made optional - but are
-  // they even tested?
-  AddStringSummary(vectors_category_sp, "", ConstString("float [4]"),
+  AddStringSummary(vectors_category_sp, "", ConstString("float[4]"),
                    vector_flags);
-  AddStringSummary(vectors_category_sp, "", ConstString("int32_t [4]"),
+  AddStringSummary(vectors_category_sp, "", ConstString("int32_t[4]"),
                    vector_flags);
-  AddStringSummary(vectors_category_sp, "", ConstString("int16_t [8]"),
+  AddStringSummary(vectors_category_sp, "", ConstString("int16_t[8]"),
                    vector_flags);
   AddStringSummary(vectors_category_sp, "", ConstString("vDouble"),
                    vector_flags);
   AddStringSummary(vectors_category_sp, "", ConstString("vFloat"),
                    vector_flags);
   AddStringSummary(vectors_category_sp, "", ConstString("vSInt8"),
                    vector_flags);
   AddStringSummary(vectors_category_sp, "", ConstString("vSInt16"),