llvm dev - Jan 2014 - [LLVMdev] RFC: Change coding standard to not indent namespaces ever

Currently there is a mixture of indented namespaces and un-indented
namespaces in both LLVM and Clang. I think this is confusing and it wastes
developer time debating the issue. I'd like to pick one and stick with it
consistently.

Indenting cannot possibly work in many contexts -- file-wide namespaces
just make no sense to indent. So I don't think we should pick "always
indent".

The common pattern to indent is when there is a small body of code which is
within a namespace. The reason for not indenting large bodies of code is
that when the namespace spans so much, the indent loses any structural
value, and turns into just a cost of losing valuable horizontal space in
which to write code. This makes sense to me.

But for small blocks of code, the indent has relatively low value -- both
the start and end are typically visible on the screen. Why bother indenting
this special case? The value seems very, very low.

So, I suggest no indent of namespaces ever. I'm happy to make the
corresponding change to the coding standards. Naturally, I'm not suggesting
rampant re-indenting of code. I'd just like to set a consistent rule going
forward so we don't debate this, and know what to do when I see a chunk of
code and about to make very significant changes to it and want to clean up
formatting while I'm there.

-Chandler
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140112/5aa74117/attachment.html>

On 01/12/2014 12:00 PM, Chandler Carruth wrote:
> Currently there is a mixture of indented namespaces and un-indented
> namespaces in both LLVM and Clang. I think this is confusing and it wastes
> developer time debating the issue. I'd like to pick one and stick with
it
> consistently.
>
> Indenting cannot possibly work in many contexts -- file-wide namespaces
> just make no sense to indent. So I don't think we should pick
"always
> indent".
>
> The common pattern to indent is when there is a small body of code which is
> within a namespace. The reason for not indenting large bodies of code is
> that when the namespace spans so much, the indent loses any structural
> value, and turns into just a cost of losing valuable horizontal space in
> which to write code. This makes sense to me.
>
> But for small blocks of code, the indent has relatively low value -- both
> the start and end are typically visible on the screen. Why bother indenting
> this special case? The value seems very, very low.
>
> So, I suggest no indent of namespaces ever. I'm happy to make the
> corresponding change to the coding standards. Naturally, I'm not
suggesting
> rampant re-indenting of code. I'd just like to set a consistent rule
going
> forward so we don't debate this, and know what to do when I see a chunk
of
> code and about to make very significant changes to it and want to clean up
> formatting while I'm there.
Hi Chandler,

thanks for pushing to further formalize our coding standards. I fully 
follow your reasoning and want to add that this is what clang-format is 
currently already doing.

Cheers,
Tobias

+1 this makes sense. It's what clang-format is already doing anyway, right?

-- Sean Silva


On Sun, Jan 12, 2014 at 6:00 AM, Chandler Carruth <chandlerc at
gmail.com>wrote:
> Currently there is a mixture of indented namespaces and un-indented
> namespaces in both LLVM and Clang. I think this is confusing and it wastes
> developer time debating the issue. I'd like to pick one and stick with
it
> consistently.
>
> Indenting cannot possibly work in many contexts -- file-wide namespaces
> just make no sense to indent. So I don't think we should pick
"always
> indent".
>
> The common pattern to indent is when there is a small body of code which
> is within a namespace. The reason for not indenting large bodies of code is
> that when the namespace spans so much, the indent loses any structural
> value, and turns into just a cost of losing valuable horizontal space in
> which to write code. This makes sense to me.
>
> But for small blocks of code, the indent has relatively low value -- both
> the start and end are typically visible on the screen. Why bother indenting
> this special case? The value seems very, very low.
>
> So, I suggest no indent of namespaces ever. I'm happy to make the
> corresponding change to the coding standards. Naturally, I'm not
suggesting
> rampant re-indenting of code. I'd just like to set a consistent rule
going
> forward so we don't debate this, and know what to do when I see a chunk
of
> code and about to make very significant changes to it and want to clean up
> formatting while I'm there.
>
> -Chandler
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140112/cb97bb80/attachment.html>

On Jan 12, 2014, at 3:00 AM, Chandler Carruth <chandlerc at gmail.com>
wrote:
> Currently there is a mixture of indented namespaces and un-indented
namespaces in both LLVM and Clang. I think this is confusing and it wastes
developer time debating the issue. I'd like to pick one and stick with it
consistently.
> 
> Indenting cannot possibly work in many contexts -- file-wide namespaces
just make no sense to indent. So I don't think we should pick "always
indent".
> 
> The common pattern to indent is when there is a small body of code which is
within a namespace. The reason for not indenting large bodies of code is that
when the namespace spans so much, the indent loses any structural value, and
turns into just a cost of losing valuable horizontal space in which to write
code. This makes sense to me.
> 
> But for small blocks of code, the indent has relatively low value -- both
the start and end are typically visible on the screen. Why bother indenting this
special case? The value seems very, very low.
> 
> So, I suggest no indent of namespaces ever. I'm happy to make the
corresponding change to the coding standards. Naturally, I'm not suggesting
rampant re-indenting of code. I'd just like to set a consistent rule going
forward so we don't debate this, and know what to do when I see a chunk of
code and about to make very significant changes to it and want to clean up
formatting while I'm there.
+1 from me.

-Chris

I think your logic is sound, +1 from me, FWIW.

~Aaron

On Mon, Jan 13, 2014 at 10:25 AM, Chris Lattner <clattner at apple.com>
wrote:
>
> On Jan 12, 2014, at 3:00 AM, Chandler Carruth <chandlerc at
gmail.com> wrote:
>
>> Currently there is a mixture of indented namespaces and un-indented
namespaces in both LLVM and Clang. I think this is confusing and it wastes
developer time debating the issue. I'd like to pick one and stick with it
consistently.
>>
>> Indenting cannot possibly work in many contexts -- file-wide namespaces
just make no sense to indent. So I don't think we should pick "always
indent".
>>
>> The common pattern to indent is when there is a small body of code
which is within a namespace. The reason for not indenting large bodies of code
is that when the namespace spans so much, the indent loses any structural value,
and turns into just a cost of losing valuable horizontal space in which to write
code. This makes sense to me.
>>
>> But for small blocks of code, the indent has relatively low value --
both the start and end are typically visible on the screen. Why bother indenting
this special case? The value seems very, very low.
>>
>> So, I suggest no indent of namespaces ever. I'm happy to make the
corresponding change to the coding standards. Naturally, I'm not suggesting
rampant re-indenting of code. I'd just like to set a consistent rule going
forward so we don't debate this, and know what to do when I see a chunk of
code and about to make very significant changes to it and want to clean up
formatting while I'm there.
>
> +1 from me.
>
> -Chris
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

LGTM.

--renato


On 12 January 2014 11:00, Chandler Carruth <chandlerc at gmail.com> wrote:
> Currently there is a mixture of indented namespaces and un-indented
> namespaces in both LLVM and Clang. I think this is confusing and it wastes
> developer time debating the issue. I'd like to pick one and stick with
it
> consistently.
>
> Indenting cannot possibly work in many contexts -- file-wide namespaces
> just make no sense to indent. So I don't think we should pick
"always
> indent".
>
> The common pattern to indent is when there is a small body of code which
> is within a namespace. The reason for not indenting large bodies of code is
> that when the namespace spans so much, the indent loses any structural
> value, and turns into just a cost of losing valuable horizontal space in
> which to write code. This makes sense to me.
>
> But for small blocks of code, the indent has relatively low value -- both
> the start and end are typically visible on the screen. Why bother indenting
> this special case? The value seems very, very low.
>
> So, I suggest no indent of namespaces ever. I'm happy to make the
> corresponding change to the coding standards. Naturally, I'm not
suggesting
> rampant re-indenting of code. I'd just like to set a consistent rule
going
> forward so we don't debate this, and know what to do when I see a chunk
of
> code and about to make very significant changes to it and want to clean up
> formatting while I'm there.
>
> -Chandler
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140113/b8159cf2/attachment.html>

Hearing no objections and support from various parts of the community,
committed in r199620. I made sure to mention using ending comments to
clarify where things end when useful.


On Sun, Jan 12, 2014 at 3:00 AM, Chandler Carruth <chandlerc at
gmail.com>wrote:
> Currently there is a mixture of indented namespaces and un-indented
> namespaces in both LLVM and Clang. I think this is confusing and it wastes
> developer time debating the issue. I'd like to pick one and stick with
it
> consistently.
>
> Indenting cannot possibly work in many contexts -- file-wide namespaces
> just make no sense to indent. So I don't think we should pick
"always
> indent".
>
> The common pattern to indent is when there is a small body of code which
> is within a namespace. The reason for not indenting large bodies of code is
> that when the namespace spans so much, the indent loses any structural
> value, and turns into just a cost of losing valuable horizontal space in
> which to write code. This makes sense to me.
>
> But for small blocks of code, the indent has relatively low value -- both
> the start and end are typically visible on the screen. Why bother indenting
> this special case? The value seems very, very low.
>
> So, I suggest no indent of namespaces ever. I'm happy to make the
> corresponding change to the coding standards. Naturally, I'm not
suggesting
> rampant re-indenting of code. I'd just like to set a consistent rule
going
> forward so we don't debate this, and know what to do when I see a chunk
of
> code and about to make very significant changes to it and want to clean up
> formatting while I'm there.
>
> -Chandler
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140119/2c78b729/attachment.html>