Re: [go-nuts] Justifying dot imports using comments

[Stefan Nilsson]

I prefer the first line to the second, even though it requires a dot import.

  *Sin*(*Pi*/2)**Sin*(*Pi*/2) + *Cos*(*Pi*)/2
  *math.Sin*(*math.Sin*/2)**math.Sin*(*math.Pi*/2) + *math.Cos*(*math.Pi*)/2

But of course in this case there is an even better solution. :)

  *0.5*

On Wednesday, April 4, 2018 at 11:30:22 PM UTC+2, rog wrote:
>
> On 2 April 2018 at 15:42, Paul Jolly  
> wrote: 
> > Speaking as someone who has written a few (albeit simple) code 
> generators 
> > that parse Go code, dot imports are a nightmare from my perspective 
> because 
> > it makes it impossible to work out from just imports alone whether an 
> > identifier belongs to the current package or not. 
> > 
> > Put another way, if you disallow dot imports, a reference to anything 
> that 
> > is outside the current package needs to be a qualified identifier, a 
> > qualified identifier that is easily determined by reference to the 
> (possibly 
> > renamed) imports in the current file. 
> > 
> > Blank imports on the other hand are only used for the side effects of 
> their 
> > inits; they do not alter the importing scope. 
> > 
> > I can't quite find the link right now, but it has long been thought/said 
> by 
> > the authors of the Go language that dot imports should not have been 
> > included in the language. 
> > 
> > All that said I suspect this was a large part of the reason that golint 
> > discourages the use of dot imports. 
>
> I agree wholeheartedly with this. In my experience they are almost 
> used for aesthetic reasons ("the package-qualified names ruin the look 
> of my DSL!") and those reasons should not trump the overall 
> readability and maintainability gains from having all externally 
> declared variables properly qualified. 
>
> If you want minimal package qualification for very frequently used 
> packages, using a shortened package identifier can work well. For 
> example, we commonly import our testing helper package with a 
> two-character package identifier because its symbols are so commonly 
> used. A single capital letter could work well too, I think. 
>
> We now also have aliases if you want to define names locally, but 
> that's an avenue I haven't explored much, and might turn out to be a 
> bad idea in the long run. 
>
> I'm rarely an absolutist when it comes to programming style, but I 
> make an exception for dot-imports. Please don't succumb to their 
> temptation. 
>

-- 
You received this message because you are subscribed to the Google Groups 
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [go-nuts] Justifying dot imports using comments

[roger peppe]

On 2 April 2018 at 15:42, Paul Jolly  wrote:
> Speaking as someone who has written a few (albeit simple) code generators
> that parse Go code, dot imports are a nightmare from my perspective because
> it makes it impossible to work out from just imports alone whether an
> identifier belongs to the current package or not.
>
> Put another way, if you disallow dot imports, a reference to anything that
> is outside the current package needs to be a qualified identifier, a
> qualified identifier that is easily determined by reference to the (possibly
> renamed) imports in the current file.
>
> Blank imports on the other hand are only used for the side effects of their
> inits; they do not alter the importing scope.
>
> I can't quite find the link right now, but it has long been thought/said by
> the authors of the Go language that dot imports should not have been
> included in the language.
>
> All that said I suspect this was a large part of the reason that golint
> discourages the use of dot imports.

I agree wholeheartedly with this. In my experience they are almost
used for aesthetic reasons ("the package-qualified names ruin the look
of my DSL!") and those reasons should not trump the overall
readability and maintainability gains from having all externally
declared variables properly qualified.

If you want minimal package qualification for very frequently used
packages, using a shortened package identifier can work well. For
example, we commonly import our testing helper package with a
two-character package identifier because its symbols are so commonly
used. A single capital letter could work well too, I think.

We now also have aliases if you want to define names locally, but
that's an avenue I haven't explored much, and might turn out to be a
bad idea in the long run.

I'm rarely an absolutist when it comes to programming style, but I
make an exception for dot-imports. Please don't succumb to their
temptation.

-- 
You received this message because you are subscribed to the Google Groups 
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [go-nuts] Justifying dot imports using comments

[Paul Jolly]

Speaking as someone who has written a few (albeit simple) code generators
that parse Go code, dot imports are a nightmare from my perspective because
it makes it impossible to work out from just imports alone whether an
identifier belongs to the current package or not.

Put another way, if you disallow dot imports, a reference to anything that
is outside the current package needs to be a qualified identifier, a
qualified identifier that is easily determined by reference to the
(possibly renamed) imports in the current file.

Blank imports on the other hand are only used for the side effects of their
inits; they do not alter the importing scope.

I can't quite find the link right now, but it has long been thought/said by
the authors of the Go language that dot imports should not have been
included in the language.

All that said I suspect this was a large part of the reason that golint
discourages the use of dot imports.



On 2 April 2018 at 15:26, Kaveh Shahbazian 
wrote:

> Why isn't it possible to justifying a dot import using a comment (like
> blank imports)?
>
> (Go 1.10.1, golint, ubuntu)
>
> --
> You received this message because you are subscribed to the Google Groups
> "golang-nuts" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to golang-nuts+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed to the Google Groups 
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.