chore: turn off the overload-adjacency rule the caches deliberately break - #164
Closed
cosmin-staicu wants to merge 1 commit into
Closed
chore: turn off the overload-adjacency rule the caches deliberately break#164cosmin-staicu wants to merge 1 commit into
cosmin-staicu wants to merge 1 commit into
Conversation
…reak S4136 fires six times across MultilayerCache, MultilayerHashCache, RedisCache and RedisHashCache, and it annotates every pull request that touches those files even though none of them introduced it. The layout it objects to is deliberate: each group of overloads sits next to the private Core helper it delegates to, so the single-key and batch groups of one name are separated by the helper that serves the first. Satisfying the rule would either move those helpers away from the overloads they support or interleave the two groups. Either reads worse than the warning. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017fwLrS3Sbcen8v6iRkUaFB Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
requested review from
alinahornet,
cosminvlad,
litheon,
lucianaparaschivei and
razvalex
as code owners
September 9, 2026 15:21
|
cosmin-staicu
added a commit
that referenced
this pull request
Sep 10, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 10, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 10, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 10, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
cosmin-staicu
added a commit
that referenced
this pull request
Sep 11, 2026
Members are laid out the way the ordering rules ask, and the rules are
on so it stays that way. The build reports no warnings at all: every
SA1201, SA1202, SA1203, SA1204, SA1210, SA1214 and IDE0040 is gone, and
S4136 with them.
What is enforced
.editorconfig already carried a curated StyleCop block, so this joins it
rather than replacing it. StyleCop.Analyzers is now referenced, and every
one of its eight categories is none except OrderingRules, with the rules
we want named individually where a category cannot reach them. Without
that the package reports around five thousand warnings.
- The OrderingRules category: order by kind, by access, constants first,
readonly first, using directives alphabetical. The category rather
than a list of ids, because it costs 28 more violations and no
maintenance.
- SA1204, static before instance, was already none further up the file.
A specific id beats a category, so the category alone would not have
applied it; it is named explicitly. SA1200 and SA1208, on using
directive placement, stay none as they were.
- SA1402, scoped to src: one top-level type per file on the shipped
surface. topLevelTypes in stylecop.json widens it past its default of
class alone. Four files held nine extra types and are now thirteen.
Tests and samples keep small helper types next to what uses them.
- SA1649 was already none and stays there: it does not recognise the OfT
suffix this repo uses for generic types, so it would rename
CacheOfT.cs and six siblings to Cache{T}.cs.
- SX1309, the alternative to SA1309, wants every field to begin with an
underscore. It belongs to no StyleCop category, so switching the
categories off does not reach it; the SX family is named off
explicitly. SA1309 is already off above for the opposite reason.
- IDE0040 was configured but silent. It is a warning now: a class member
always states its accessibility, an interface member never repeats the
public it already has. 25 interface members carried a redundant public
and two class members had none.
How the members were moved
Neither dotnet format nor the Roslynator CLI can drive StyleCop's fixer
for the ordering rules — both answer that no code fix was found — and
Rider's layout engine ranks constants and statics above accessibility,
which pushes SA1202 up rather than down. So a Roslyn pass did it: parse,
sort each type's members by StyleCop's default elementOrder, write the
tree back, which carries each member's doc comments and blank lines with
it.
Three rules kept it safe. The sort is stable, so anything the comparer
calls equal keeps the order the author chose. A field whose initializer
reads a sibling member, or this, is pinned where it is, because C# runs
field initializers in textual order. And a file containing #region or
#if is refused outright, since those directives are trivia on the
members around them and a sort can carry one away from its partner.
That left seven conditional-compilation files and six other places to
settle by hand: three field blocks moved as units so their initializers
keep reading what they read, an interface that sat below a class in its
file, and the two buffer tests whose nested helper lives beside the
tests that use it.
Also here
S4136 had been reporting eight non-adjacent overload groups on every
build, and #164 proposed turning the rule off. Measuring first said
otherwise: every group is of uniform accessibility, five are split by a
single private Core helper sitting between the public overloads it
serves, and moving that helper below the group is what SA1202 asks for
anyway. The two rules agree rather than conflict.
KeyMasker.IsMasking carried two summary tags, having taken ShouldMask's
when it was inserted above it. ShouldMask has its documentation back.
Verification
The reordering is a move and nothing else, which the diff shows: the
insertions and deletions match. Both suites are unchanged at 1701 on
net10 and 1680 on net8 — the count is the check that matters for the
conditional files, because a test carried out of a NET9_0_OR_GREATER
block would raise the net8 total and one carried in would lower it. #if
pairs stay balanced. The packed nuspec is untouched.
Claude-Session: https://claude.ai/code/session_01NsPw6MZHPGmpbo6WDuLzF1
Signed-off-by: Cosmin Staicu <cosmin.staicu@uipath.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
S4136("All X method overloads should be adjacent") fires six times, inMultilayerCache,MultilayerHashCache,RedisCacheandRedisHashCache. None of them is new. Because GitHub annotates warnings on changed files, every pull request that touches one of those files carries them, which is where this came from: PR #160 shows all six and introduced none.Why not reorder instead
The layout the rule objects to is deliberate. Each group of overloads is followed by the private
Corehelper it delegates to:Making the two groups adjacent means either moving the helpers away from the overloads they serve, or interleaving single-key and batch code. Both read worse than the warning, so the rule is switched off with that reason recorded next to it, in the same style as the other rules the repository already disables.
Verification
Solution builds clean. Behavior unchanged: this is one
.editorconfigentry. CI on this pull request is the check that the six annotations are gone.🤖 Generated with Claude Code
https://claude.ai/code/session_017fwLrS3Sbcen8v6iRkUaFB