| Message ID | 20250205231208.1480762-29-jsnow@redhat.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | docs: add sphinx-domain rST generator to qapidoc | expand |
John Snow <jsnow@redhat.com> writes: > This is being done primarily to ensure consistency between the source > documents and the final, rendered HTML output. Because > member/feature/returns sections will always appear in a visually grouped > element in the HTML output, prohibiting free paragraphs between those > sections ensures ordering consistency between source and the final > render. > > Additionally, prohibiting such "middle" text paragraphs allows us to > classify all plain text sections as either "intro" or "detail" > sections, because these sections must either appear before structured > elements ("intro") or afterwards ("detail"). > > This keeps the inlining algorithm simpler with fewer "splice" points > when inlining multiple documentation blocks. Mention the two "middle" paragraphs you have to eliminate in this patch? > > Signed-off-by: John Snow <jsnow@redhat.com> > --- > qapi/net.json | 4 ++-- > qapi/qom.json | 4 ++-- > scripts/qapi/parser.py | 16 ++++++++++++++++ > tests/qapi-schema/doc-good.json | 4 ++-- > tests/qapi-schema/doc-good.out | 4 ++-- > tests/qapi-schema/doc-good.txt | 8 ++++---- > 6 files changed, 28 insertions(+), 12 deletions(-) > > diff --git a/qapi/net.json b/qapi/net.json > index 2739a2f4233..49bc7de64e9 100644 > --- a/qapi/net.json > +++ b/qapi/net.json > @@ -655,13 +655,13 @@ > # this to zero disables this function. This member is mutually > # exclusive with @reconnect. (default: 0) (Since: 9.2) > # > -# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. > -# > # Features: > # > # @deprecated: Member @reconnect is deprecated. Use @reconnect-ms > # instead. > # > +# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. > +# > # Since: 7.2 > ## > { 'struct': 'NetdevStreamOptions', The text moved applies to member @addr. You're moving it even farther away from @addr. Move it into @addr instead? Could be done as a separate cleanup patch to keep this one as simple as possible; matter of taste. The same text is in NetdevDgramOptions below, where it applies to both @remote and @local. It just happens to follow @remote and @local immediately, because there are no other members and no features. Hmm. Ideally, we'd have a way to put such notes next to the stuff they apply to without having to rely on happy accidents like "no features". Alternatively, have a way to link stuff and note. Footnotes? Food for thought, not demand. > diff --git a/qapi/qom.json b/qapi/qom.json > index 28ce24cd8d0..11277d1f84c 100644 > --- a/qapi/qom.json > +++ b/qapi/qom.json > @@ -195,12 +195,12 @@ > # > # @typename: the type name of an object > # > +# Returns: a list of ObjectPropertyInfo describing object properties > +# > # .. note:: Objects can create properties at runtime, for example to > # describe links between different devices and/or objects. These > # properties are not included in the output of this command. > # > -# Returns: a list of ObjectPropertyInfo describing object properties > -# > # Since: 2.12 > ## This move is fine. Placing notes at the end is more common already. > { 'command': 'qom-list-properties', > diff --git a/scripts/qapi/parser.py b/scripts/qapi/parser.py > index b2f77ffdd7a..c5d2b950a82 100644 > --- a/scripts/qapi/parser.py > +++ b/scripts/qapi/parser.py > @@ -500,6 +500,20 @@ def get_doc(self) -> 'QAPIDoc': > self.accept(False) > line = self.get_doc_line() > have_tagged = False > + no_more_tags = False > + > + def _tag_check(what: str) -> None: > + if what in ('TODO', 'Since'): > + return > + > + if no_more_tags: > + raise QAPIParseError( > + self, > + f"{what!r} section cannot appear after free " > + "paragraphs that follow other tagged sections. " > + "Move this section upwards with the preceding " > + "tagged sections." > + ) Why !r conversion? Error messages should be a single, short phrase, no punctuation at the end. Sometimes a helpful hint is desirable. Here's one in expr.py: raise QAPISemError( info, "%s has unknown key%s %s\nValid keys are %s." % (source, 's' if len(unknown) > 1 else '', pprint(unknown), pprint(allowed))) Needs a negative test case. Aside: we should probably convert most string interpolation to f-strings en masse at some point. > > while line is not None: > # Blank lines > @@ -513,6 +527,7 @@ def get_doc(self) -> 'QAPIDoc': > if doc.features: > raise QAPIParseError( > self, "duplicated 'Features:' line") > + _tag_check("Features") > self.accept(False) > line = self.get_doc_line() > while line == '': > @@ -576,6 +591,7 @@ def get_doc(self) -> 'QAPIDoc': > ) > raise QAPIParseError(self, emsg) > > + _tag_check(match.group(1)) > doc.new_tagged_section( > self.info, > QAPIDoc.Kind.from_string(match.group(1)) > diff --git a/tests/qapi-schema/doc-good.json b/tests/qapi-schema/doc-good.json > index f64bf38d854..14b2091b08f 100644 > --- a/tests/qapi-schema/doc-good.json > +++ b/tests/qapi-schema/doc-good.json > @@ -157,12 +157,12 @@ > # @cmd-feat1: a feature > # @cmd-feat2: another feature > # > -# .. note:: @arg3 is undocumented > -# > # Returns: @Object > # > # Errors: some > # > +# .. note:: @arg3 is undocumented > +# This used to be right next to @arg1 and arg2 (commit 80d1f2e4a5d) until commit 79598c8a634 added features in between. This patch adds more stuff there. Right next is clearly the best spot, but this is just a test, so it doesn't really matter. Related: NetdevDgramOptions' note discussed above. > # TODO: frobnicate > # > # .. admonition:: Notes [...]
John Snow <jsnow@redhat.com> writes: > This is being done primarily to ensure consistency between the source > documents and the final, rendered HTML output. Because > member/feature/returns sections will always appear in a visually grouped > element in the HTML output, prohibiting free paragraphs between those > sections ensures ordering consistency between source and the final > render. > > Additionally, prohibiting such "middle" text paragraphs allows us to > classify all plain text sections as either "intro" or "detail" > sections, because these sections must either appear before structured > elements ("intro") or afterwards ("detail"). > > This keeps the inlining algorithm simpler with fewer "splice" points > when inlining multiple documentation blocks. > > Signed-off-by: John Snow <jsnow@redhat.com> [...] > diff --git a/scripts/qapi/parser.py b/scripts/qapi/parser.py > index b2f77ffdd7a..c5d2b950a82 100644 > --- a/scripts/qapi/parser.py > +++ b/scripts/qapi/parser.py > @@ -500,6 +500,20 @@ def get_doc(self) -> 'QAPIDoc': > self.accept(False) > line = self.get_doc_line() > have_tagged = False > + no_more_tags = False > + > + def _tag_check(what: str) -> None: > + if what in ('TODO', 'Since'): > + return > + > + if no_more_tags: > + raise QAPIParseError( > + self, > + f"{what!r} section cannot appear after free " > + "paragraphs that follow other tagged sections. " > + "Move this section upwards with the preceding " > + "tagged sections." > + ) Negative test case(s), please. > > while line is not None: > # Blank lines > @@ -513,6 +527,7 @@ def get_doc(self) -> 'QAPIDoc': > if doc.features: > raise QAPIParseError( > self, "duplicated 'Features:' line") > + _tag_check("Features") > self.accept(False) > line = self.get_doc_line() > while line == '': > @@ -576,6 +591,7 @@ def get_doc(self) -> 'QAPIDoc': > ) > raise QAPIParseError(self, emsg) > > + _tag_check(match.group(1)) > doc.new_tagged_section( > self.info, > QAPIDoc.Kind.from_string(match.group(1)) [...]
On Wed, Feb 12, 2025 at 4:06 AM Markus Armbruster <armbru@redhat.com> wrote: > John Snow <jsnow@redhat.com> writes: > > > This is being done primarily to ensure consistency between the source > > documents and the final, rendered HTML output. Because > > member/feature/returns sections will always appear in a visually grouped > > element in the HTML output, prohibiting free paragraphs between those > > sections ensures ordering consistency between source and the final > > render. > > > > Additionally, prohibiting such "middle" text paragraphs allows us to > > classify all plain text sections as either "intro" or "detail" > > sections, because these sections must either appear before structured > > elements ("intro") or afterwards ("detail"). > > > > This keeps the inlining algorithm simpler with fewer "splice" points > > when inlining multiple documentation blocks. > > Mention the two "middle" paragraphs you have to eliminate in this patch? > OK; I will mention that this patch adjusts the source documentation but I won't go into detail on which. You can read the patch to find out easily enough. > > > > > Signed-off-by: John Snow <jsnow@redhat.com> > > --- > > qapi/net.json | 4 ++-- > > qapi/qom.json | 4 ++-- > > scripts/qapi/parser.py | 16 ++++++++++++++++ > > tests/qapi-schema/doc-good.json | 4 ++-- > > tests/qapi-schema/doc-good.out | 4 ++-- > > tests/qapi-schema/doc-good.txt | 8 ++++---- > > 6 files changed, 28 insertions(+), 12 deletions(-) > > > > diff --git a/qapi/net.json b/qapi/net.json > > index 2739a2f4233..49bc7de64e9 100644 > > --- a/qapi/net.json > > +++ b/qapi/net.json > > @@ -655,13 +655,13 @@ > > # this to zero disables this function. This member is mutually > > # exclusive with @reconnect. (default: 0) (Since: 9.2) > > # > > -# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. > > -# > > # Features: > > # > > # @deprecated: Member @reconnect is deprecated. Use @reconnect-ms > > # instead. > > # > > +# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. > > +# > > # Since: 7.2 > > ## > > { 'struct': 'NetdevStreamOptions', > > The text moved applies to member @addr. You're moving it even farther > away from @addr. Move it into @addr instead? Could be done as a > separate cleanup patch to keep this one as simple as possible; matter of > taste. > Mmm, I was doing a mechanical hacksaw job here, admittedly. I can do a more tactful adjustment. I think it should be in this patch in order to preserve the context of precisely *why* it was juggled around, because I admit in this one case it is a slight downgrade. Moving it into @addr. > > The same text is in NetdevDgramOptions below, where it applies to both > @remote and @local. It just happens to follow @remote and @local > immediately, because there are no other members and no features. Hmm. > > Ideally, we'd have a way to put such notes next to the stuff they apply > to without having to rely on happy accidents like "no features". > Alternatively, have a way to link stuff and note. Footnotes? Food for > thought, not demand. > Yes, we discussed this at KVM Forum and I was dreaming of some complicated solution like "section-details: " or something that allows us to add amendments to documentation regions that aren't associated with any one particular member or feature, but can be inserted visually at that point. I know it's a capability you'd like to preserve, but I think we only use it once, so I'd be happy to push this off until a bit later and just suffer the indignity of slightly suboptimal documentation in one spot until then in exchange for the massive upgrade everywhere else. What would help a good deal is if you could brainstorm some source syntax that you think would be pleasant for the purpose, and then for my end I can worry about how to munge the docutils tree and HTML renderer to make it happen in some pleasing way. For now... "Figure out how to add notes or footnotes to the members section as a whole" added to the "for later" part of my tasklist? > > > diff --git a/qapi/qom.json b/qapi/qom.json > > index 28ce24cd8d0..11277d1f84c 100644 > > --- a/qapi/qom.json > > +++ b/qapi/qom.json > > @@ -195,12 +195,12 @@ > > # > > # @typename: the type name of an object > > # > > +# Returns: a list of ObjectPropertyInfo describing object properties > > +# > > # .. note:: Objects can create properties at runtime, for example to > > # describe links between different devices and/or objects. These > > # properties are not included in the output of this command. > > # > > -# Returns: a list of ObjectPropertyInfo describing object properties > > -# > > # Since: 2.12 > > ## > > This move is fine. Placing notes at the end is more common already. > > > { 'command': 'qom-list-properties', > > diff --git a/scripts/qapi/parser.py b/scripts/qapi/parser.py > > index b2f77ffdd7a..c5d2b950a82 100644 > > --- a/scripts/qapi/parser.py > > +++ b/scripts/qapi/parser.py > > @@ -500,6 +500,20 @@ def get_doc(self) -> 'QAPIDoc': > > self.accept(False) > > line = self.get_doc_line() > > have_tagged = False > > + no_more_tags = False > > + > > + def _tag_check(what: str) -> None: > > + if what in ('TODO', 'Since'): > > + return > > + > > + if no_more_tags: > > + raise QAPIParseError( > > + self, > > + f"{what!r} section cannot appear after free " > > + "paragraphs that follow other tagged sections. " > > + "Move this section upwards with the preceding " > > + "tagged sections." > > + ) > > Why !r conversion? > It just adds quotes so it'll print like: "Returns" section cannot appear after ... I thought it looked nicer codewise than: f'"{what}" section cannot appear after ...' > Error messages should be a single, short phrase, no punctuation at the > end. Sometimes a helpful hint is desirable. Here's one in expr.py: > > raise QAPISemError( > info, > "%s has unknown key%s %s\nValid keys are %s." > % (source, 's' if len(unknown) > 1 else '', > pprint(unknown), pprint(allowed))) > Cold day in hell when I successfully remember to omit the punctuation. Will rephrase and fix. > > Needs a negative test case. > Yes, I didn't add any new tests because I was being lazy and wanted to see which things you'd toss out before I had to write them. No reason to do all that work in advance of review. Please ask for tests wherever you feel they are required and I'll add them. In this case, I knew you had some qualms about this patch, so I was hesitant ... > > Aside: we should probably convert most string interpolation to f-strings > en masse at some point. > Yeah, just putting it off because the review for that sounds like a hassle ;) I can do a mechanical conversion to get you started and then let you finesse it if you want? We can share the pain. If you really wanna have a flag day about it, we can just run the black formatter on everything and get it all over with at once... later stuff, to be clear. > > > > > while line is not None: > > # Blank lines > > @@ -513,6 +527,7 @@ def get_doc(self) -> 'QAPIDoc': > > if doc.features: > > raise QAPIParseError( > > self, "duplicated 'Features:' line") > > + _tag_check("Features") > > self.accept(False) > > line = self.get_doc_line() > > while line == '': > > @@ -576,6 +591,7 @@ def get_doc(self) -> 'QAPIDoc': > > ) > > raise QAPIParseError(self, emsg) > > > > + _tag_check(match.group(1)) > > doc.new_tagged_section( > > self.info, > > QAPIDoc.Kind.from_string(match.group(1)) > > diff --git a/tests/qapi-schema/doc-good.json > b/tests/qapi-schema/doc-good.json > > index f64bf38d854..14b2091b08f 100644 > > --- a/tests/qapi-schema/doc-good.json > > +++ b/tests/qapi-schema/doc-good.json > > @@ -157,12 +157,12 @@ > > # @cmd-feat1: a feature > > # @cmd-feat2: another feature > > # > > -# .. note:: @arg3 is undocumented > > -# > > # Returns: @Object > > # > > # Errors: some > > # > > +# .. note:: @arg3 is undocumented > > +# > > This used to be right next to @arg1 and arg2 (commit 80d1f2e4a5d) until > commit 79598c8a634 added features in between. This patch adds more > stuff there. Right next is clearly the best spot, but this is just a > test, so it doesn't really matter. Related: NetdevDgramOptions' note > discussed above. > So long as it doesn't disturb the point of what is being tested. It's not always directly clear when adjusting the good doc what precisely is being tested. > > > # TODO: frobnicate > > # > > # .. admonition:: Notes > > [...] > >
John Snow <jsnow@redhat.com> writes: > On Wed, Feb 12, 2025 at 4:06 AM Markus Armbruster <armbru@redhat.com> wrote: > >> John Snow <jsnow@redhat.com> writes: >> >> > This is being done primarily to ensure consistency between the source >> > documents and the final, rendered HTML output. Because >> > member/feature/returns sections will always appear in a visually grouped >> > element in the HTML output, prohibiting free paragraphs between those >> > sections ensures ordering consistency between source and the final >> > render. >> > >> > Additionally, prohibiting such "middle" text paragraphs allows us to >> > classify all plain text sections as either "intro" or "detail" >> > sections, because these sections must either appear before structured >> > elements ("intro") or afterwards ("detail"). >> > >> > This keeps the inlining algorithm simpler with fewer "splice" points >> > when inlining multiple documentation blocks. >> >> Mention the two "middle" paragraphs you have to eliminate in this patch? >> > > OK; I will mention that this patch adjusts the source documentation but I > won't go into detail on which. You can read the patch to find out easily > enough. > > >> >> > >> > Signed-off-by: John Snow <jsnow@redhat.com> >> > --- >> > qapi/net.json | 4 ++-- >> > qapi/qom.json | 4 ++-- >> > scripts/qapi/parser.py | 16 ++++++++++++++++ >> > tests/qapi-schema/doc-good.json | 4 ++-- >> > tests/qapi-schema/doc-good.out | 4 ++-- >> > tests/qapi-schema/doc-good.txt | 8 ++++---- >> > 6 files changed, 28 insertions(+), 12 deletions(-) >> > >> > diff --git a/qapi/net.json b/qapi/net.json >> > index 2739a2f4233..49bc7de64e9 100644 >> > --- a/qapi/net.json >> > +++ b/qapi/net.json >> > @@ -655,13 +655,13 @@ >> > # this to zero disables this function. This member is mutually >> > # exclusive with @reconnect. (default: 0) (Since: 9.2) >> > # >> > -# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. >> > -# >> > # Features: >> > # >> > # @deprecated: Member @reconnect is deprecated. Use @reconnect-ms >> > # instead. >> > # >> > +# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. >> > +# >> > # Since: 7.2 >> > ## >> > { 'struct': 'NetdevStreamOptions', >> >> The text moved applies to member @addr. You're moving it even farther >> away from @addr. Move it into @addr instead? Could be done as a >> separate cleanup patch to keep this one as simple as possible; matter of >> taste. >> > > Mmm, I was doing a mechanical hacksaw job here, admittedly. I can do a more > tactful adjustment. I think it should be in this patch in order to preserve > the context of precisely *why* it was juggled around, because I admit in > this one case it is a slight downgrade. > > Moving it into @addr. > > >> >> The same text is in NetdevDgramOptions below, where it applies to both >> @remote and @local. It just happens to follow @remote and @local >> immediately, because there are no other members and no features. Hmm. >> >> Ideally, we'd have a way to put such notes next to the stuff they apply >> to without having to rely on happy accidents like "no features". >> Alternatively, have a way to link stuff and note. Footnotes? Food for >> thought, not demand. >> > > Yes, we discussed this at KVM Forum and I was dreaming of some complicated > solution like "section-details: " or something that allows us to add > amendments to documentation regions that aren't associated with any one > particular member or feature, but can be inserted visually at that point. > > I know it's a capability you'd like to preserve, but I think we only use it > once, so I'd be happy to push this off until a bit later and just suffer > the indignity of slightly suboptimal documentation in one spot until then > in exchange for the massive upgrade everywhere else. If minor degradations are the price of major improvement, we pay. > What would help a good deal is if you could brainstorm some source syntax > that you think would be pleasant for the purpose, and then for my end I can > worry about how to munge the docutils tree and HTML renderer to make it > happen in some pleasing way. Can we make do with just ReST? Footnotes? Best if we can control their placement somehow. > For now... "Figure out how to add notes or footnotes to the members section > as a whole" added to the "for later" part of my tasklist? Yes, please, with the understanding that this is *not* a blocker for getting the new doc generator accepted. >> > diff --git a/qapi/qom.json b/qapi/qom.json >> > index 28ce24cd8d0..11277d1f84c 100644 >> > --- a/qapi/qom.json >> > +++ b/qapi/qom.json >> > @@ -195,12 +195,12 @@ >> > # >> > # @typename: the type name of an object >> > # >> > +# Returns: a list of ObjectPropertyInfo describing object properties >> > +# >> > # .. note:: Objects can create properties at runtime, for example to >> > # describe links between different devices and/or objects. These >> > # properties are not included in the output of this command. >> > # >> > -# Returns: a list of ObjectPropertyInfo describing object properties >> > -# >> > # Since: 2.12 >> > ## >> >> This move is fine. Placing notes at the end is more common already. >> >> > { 'command': 'qom-list-properties', >> > diff --git a/scripts/qapi/parser.py b/scripts/qapi/parser.py >> > index b2f77ffdd7a..c5d2b950a82 100644 >> > --- a/scripts/qapi/parser.py >> > +++ b/scripts/qapi/parser.py >> > @@ -500,6 +500,20 @@ def get_doc(self) -> 'QAPIDoc': >> > self.accept(False) >> > line = self.get_doc_line() >> > have_tagged = False >> > + no_more_tags = False >> > + >> > + def _tag_check(what: str) -> None: >> > + if what in ('TODO', 'Since'): >> > + return >> > + >> > + if no_more_tags: >> > + raise QAPIParseError( >> > + self, >> > + f"{what!r} section cannot appear after free " >> > + "paragraphs that follow other tagged sections. " >> > + "Move this section upwards with the preceding " >> > + "tagged sections." >> > + ) >> >> Why !r conversion? >> > > It just adds quotes so it'll print like: "Returns" section cannot appear > after ... > I thought it looked nicer codewise than: f'"{what}" section cannot appear > after ...' Prior art: raise QAPISemError( info, "duplicated '%s' section" % kind) raise QAPISemError( self.returns.info, "'Returns' section, but command doesn't return anything") raise QAPISemError( self.returns.info, "'Returns' section is only valid for commands") raise QAPISemError( self.errors.info, "'Errors' section is only valid for commands") Let's stick to this style. >> Error messages should be a single, short phrase, no punctuation at the >> end. Sometimes a helpful hint is desirable. Here's one in expr.py: >> > >> raise QAPISemError( >> info, >> "%s has unknown key%s %s\nValid keys are %s." >> % (source, 's' if len(unknown) > 1 else '', >> pprint(unknown), pprint(allowed))) >> > > Cold day in hell when I successfully remember to omit the punctuation. Will > rephrase and fix. :) >> Needs a negative test case. >> > > Yes, I didn't add any new tests because I was being lazy and wanted to see > which things you'd toss out before I had to write them. No reason to do all > that work in advance of review. Please ask for tests wherever you feel they > are required and I'll add them. In this case, I knew you had some qualms > about this patch, so I was hesitant ... Good strategy. I'd appreciate notes on known gaps in the commit message and/or the cover letter, though. Don't go out of your way to find gaps, just make a reasonable effort to write down the ones you already know. >> Aside: we should probably convert most string interpolation to f-strings >> en masse at some point. >> > > Yeah, just putting it off because the review for that sounds like a hassle > ;) > I can do a mechanical conversion to get you started and then let you > finesse it if you want? > We can share the pain. > > If you really wanna have a flag day about it, we can just run the black > formatter on everything and get it all over with at once... > > later stuff, to be clear. Yes, all of it. >> > >> > while line is not None: >> > # Blank lines >> > @@ -513,6 +527,7 @@ def get_doc(self) -> 'QAPIDoc': >> > if doc.features: >> > raise QAPIParseError( >> > self, "duplicated 'Features:' line") >> > + _tag_check("Features") >> > self.accept(False) >> > line = self.get_doc_line() >> > while line == '': >> > @@ -576,6 +591,7 @@ def get_doc(self) -> 'QAPIDoc': >> > ) >> > raise QAPIParseError(self, emsg) >> > >> > + _tag_check(match.group(1)) >> > doc.new_tagged_section( >> > self.info, >> > QAPIDoc.Kind.from_string(match.group(1)) >> > diff --git a/tests/qapi-schema/doc-good.json b/tests/qapi-schema/doc-good.json >> > index f64bf38d854..14b2091b08f 100644 >> > --- a/tests/qapi-schema/doc-good.json >> > +++ b/tests/qapi-schema/doc-good.json >> > @@ -157,12 +157,12 @@ >> > # @cmd-feat1: a feature >> > # @cmd-feat2: another feature >> > # >> > -# .. note:: @arg3 is undocumented >> > -# >> > # Returns: @Object >> > # >> > # Errors: some >> > # >> > +# .. note:: @arg3 is undocumented >> > +# >> >> This used to be right next to @arg1 and arg2 (commit 80d1f2e4a5d) until >> commit 79598c8a634 added features in between. This patch adds more >> stuff there. Right next is clearly the best spot, but this is just a >> test, so it doesn't really matter. Related: NetdevDgramOptions' note >> discussed above. >> > > So long as it doesn't disturb the point of what is being tested. It's not > always directly clear when adjusting the good doc what precisely is being > tested. True. >> > # TODO: frobnicate >> > # >> > # .. admonition:: Notes >> >> [...] >> >>
diff --git a/qapi/net.json b/qapi/net.json index 2739a2f4233..49bc7de64e9 100644 --- a/qapi/net.json +++ b/qapi/net.json @@ -655,13 +655,13 @@ # this to zero disables this function. This member is mutually # exclusive with @reconnect. (default: 0) (Since: 9.2) # -# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. -# # Features: # # @deprecated: Member @reconnect is deprecated. Use @reconnect-ms # instead. # +# Only SocketAddress types 'unix', 'inet' and 'fd' are supported. +# # Since: 7.2 ## { 'struct': 'NetdevStreamOptions', diff --git a/qapi/qom.json b/qapi/qom.json index 28ce24cd8d0..11277d1f84c 100644 --- a/qapi/qom.json +++ b/qapi/qom.json @@ -195,12 +195,12 @@ # # @typename: the type name of an object # +# Returns: a list of ObjectPropertyInfo describing object properties +# # .. note:: Objects can create properties at runtime, for example to # describe links between different devices and/or objects. These # properties are not included in the output of this command. # -# Returns: a list of ObjectPropertyInfo describing object properties -# # Since: 2.12 ## { 'command': 'qom-list-properties', diff --git a/scripts/qapi/parser.py b/scripts/qapi/parser.py index b2f77ffdd7a..c5d2b950a82 100644 --- a/scripts/qapi/parser.py +++ b/scripts/qapi/parser.py @@ -500,6 +500,20 @@ def get_doc(self) -> 'QAPIDoc': self.accept(False) line = self.get_doc_line() have_tagged = False + no_more_tags = False + + def _tag_check(what: str) -> None: + if what in ('TODO', 'Since'): + return + + if no_more_tags: + raise QAPIParseError( + self, + f"{what!r} section cannot appear after free " + "paragraphs that follow other tagged sections. " + "Move this section upwards with the preceding " + "tagged sections." + ) while line is not None: # Blank lines @@ -513,6 +527,7 @@ def get_doc(self) -> 'QAPIDoc': if doc.features: raise QAPIParseError( self, "duplicated 'Features:' line") + _tag_check("Features") self.accept(False) line = self.get_doc_line() while line == '': @@ -576,6 +591,7 @@ def get_doc(self) -> 'QAPIDoc': ) raise QAPIParseError(self, emsg) + _tag_check(match.group(1)) doc.new_tagged_section( self.info, QAPIDoc.Kind.from_string(match.group(1)) diff --git a/tests/qapi-schema/doc-good.json b/tests/qapi-schema/doc-good.json index f64bf38d854..14b2091b08f 100644 --- a/tests/qapi-schema/doc-good.json +++ b/tests/qapi-schema/doc-good.json @@ -157,12 +157,12 @@ # @cmd-feat1: a feature # @cmd-feat2: another feature # -# .. note:: @arg3 is undocumented -# # Returns: @Object # # Errors: some # +# .. note:: @arg3 is undocumented +# # TODO: frobnicate # # .. admonition:: Notes diff --git a/tests/qapi-schema/doc-good.out b/tests/qapi-schema/doc-good.out index 3eb28503f6b..77edadb6e40 100644 --- a/tests/qapi-schema/doc-good.out +++ b/tests/qapi-schema/doc-good.out @@ -168,12 +168,12 @@ description starts on the same line a feature feature=cmd-feat2 another feature - section=Detail -.. note:: @arg3 is undocumented section=Returns @Object section=Errors some + section=Detail +.. note:: @arg3 is undocumented section=Todo frobnicate section=Detail diff --git a/tests/qapi-schema/doc-good.txt b/tests/qapi-schema/doc-good.txt index cb37db606a6..dc40b3f7046 100644 --- a/tests/qapi-schema/doc-good.txt +++ b/tests/qapi-schema/doc-good.txt @@ -193,10 +193,6 @@ Features "cmd-feat2" another feature -Note: - - "arg3" is undocumented - Returns ~~~~~~~ @@ -209,6 +205,10 @@ Errors some +Note: + + "arg3" is undocumented + Notes: * Lorem ipsum dolor sit amet
This is being done primarily to ensure consistency between the source documents and the final, rendered HTML output. Because member/feature/returns sections will always appear in a visually grouped element in the HTML output, prohibiting free paragraphs between those sections ensures ordering consistency between source and the final render. Additionally, prohibiting such "middle" text paragraphs allows us to classify all plain text sections as either "intro" or "detail" sections, because these sections must either appear before structured elements ("intro") or afterwards ("detail"). This keeps the inlining algorithm simpler with fewer "splice" points when inlining multiple documentation blocks. Signed-off-by: John Snow <jsnow@redhat.com> --- qapi/net.json | 4 ++-- qapi/qom.json | 4 ++-- scripts/qapi/parser.py | 16 ++++++++++++++++ tests/qapi-schema/doc-good.json | 4 ++-- tests/qapi-schema/doc-good.out | 4 ++-- tests/qapi-schema/doc-good.txt | 8 ++++---- 6 files changed, 28 insertions(+), 12 deletions(-)