www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - VSCode plugins

reply Manu via Digitalmars-d <digitalmars-d puremagic.com> writes:
I've been using code-d for a while, and it generally works well. I've also
noticed there's another plugin available, which at the time I first looked,
appeared to be older and less-featured than code-d.

I've recently had a couple of colleagues ask me which plugin to install,
and I noticed that both seem to be up-to-date these days, and this leads to
confusion.

Looking at the feature list, it appears that both plugins do mostly the
same stuff.
My feeling is, having 2 very similar plugins is confusing to potential
users, and it undermines user confidence. Ie, users have the feeling that
they're competing hacky things maintained by some guy, rather than
something that's more 'official' with consolidated community support. I
also tend to presume in these situations that the 'proper' one is the one
with the most users/installs, but that's not clear either in this case.
I know this has nothing to do with the truth, but it's about perception and
first impressions. Little things matter.

If authors of both plugins are active here, I ask; why have 2 separate
plugins?
I can't imagine any reason for divergence. I would be a lot more
comfortable if there was only one with multiple contributors. Projects with
many contributors always inspire a lot more confidence than multiple
overlapping projects with one contributor each...

So, is there a reason not to merge the projects beyond ego?
Jul 03 2017
next sibling parent reply bauss <jj_1337 live.dk> writes:
On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:

 If authors of both plugins are active here, I ask; why have 2 
 separate


The same reason there are multiple editors.
Jul 03 2017
parent reply Manu via Digitalmars-d <digitalmars-d puremagic.com> writes:
On 4 July 2017 at 14:59, bauss via Digitalmars-d <
digitalmars-d puremagic.com> wrote:


 On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:


 If authors of both plugins are active here, I ask; why have 2 separate


 The same reason there are multiple editors.


Errr. No.
I don't think you can make an argument that one plugin performs in a way
that conforms to some peoples taste more than others... they do exactly the
same things, and even in the same way using the same tooling.
There might be implementation quality differences, I don't know, but that
leads to the conclusion that merging them would produce the best quality
result.
Jul 03 2017
parent Jerry <hurricane hereiam.com> writes:
On Tuesday, 4 July 2017 at 06:26:52 UTC, Manu wrote:

 On 4 July 2017 at 14:59, bauss via Digitalmars-d < 
 digitalmars-d puremagic.com> wrote:


 On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:


 If authors of both plugins are active here, I ask; why have 2 
 separate


 The same reason there are multiple editors.


 Errr. No.
 I don't think you can make an argument that one plugin performs 
 in a way
 that conforms to some peoples taste more than others... they do 
 exactly the
 same things, and even in the same way using the same tooling.
 There might be implementation quality differences, I don't 
 know, but that
 leads to the conclusion that merging them would produce the 
 best quality
 result.


Personally I use dlang-vscode because it doesn't tak on an extra 
process used for communication. I can understand wanting to put 
common code into one location so that multiple editors can reuse 
it, but ultimately you end up writing the same code for the 
editors anyways. Just now instead of communicating with 
DCD/DScanner/etc it is doing the same thing just communicating 
with this new process now. It's also really buggy and you don't 
get any information when something goes wrong. When it does go 
wrong in code-d, the workspace-d crashes and you now lose access 
to every feature. Where as in dlang-vscode if something goes 
wrong with DCD you still have access to features covered by the 
other processes. So they are different, and I do have my 
preference after using both of them.
Jul 14 2017
prev sibling next sibling parent reply Meta <jared771 gmail.com> writes:
On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:

 I've been using code-d for a while, and it generally works 
 well. I've also noticed there's another plugin available, which 
 at the time I first looked, appeared to be older and 
 less-featured than code-d.

 I've recently had a couple of colleagues ask me which plugin to 
 install, and I noticed that both seem to be up-to-date these 
 days, and this leads to confusion.

 Looking at the feature list, it appears that both plugins do 
 mostly the
 same stuff.
 My feeling is, having 2 very similar plugins is confusing to 
 potential
 users, and it undermines user confidence. Ie, users have the 
 feeling that
 they're competing hacky things maintained by some guy, rather 
 than
 something that's more 'official' with consolidated community 
 support. I
 also tend to presume in these situations that the 'proper' one 
 is the one
 with the most users/installs, but that's not clear either in 
 this case.
 I know this has nothing to do with the truth, but it's about 
 perception and
 first impressions. Little things matter.

 If authors of both plugins are active here, I ask; why have 2 
 separate
 plugins?
 I can't imagine any reason for divergence. I would be a lot more
 comfortable if there was only one with multiple contributors. 
 Projects with
 many contributors always inspire a lot more confidence than 
 multiple
 overlapping projects with one contributor each...

 So, is there a reason not to merge the projects beyond ego?


It's just different users developing different solutions. I 
disagree with the notion that having multiple competing, up to 
date implementations would "undermine user confidence" in D. 
Quite the opposite, I'd think.
Jul 04 2017
parent Manu via Digitalmars-d <digitalmars-d puremagic.com> writes:
On 4 July 2017 at 22:41, Meta via Digitalmars-d <digitalmars-d puremagic.com

 wrote:



 It's just different users developing different solutions. I disagree with


 the notion that having multiple competing, up to date implementations would
 "undermine user confidence" in D. Quite the opposite, I'd think.


My point was, I feel like this is a case of different users developing the
same solution though.
Sure, it's reasonable to disagree, but I raise it because I have multiple
counts of anecdotal evidence from prospective users that it is a problem.

Let's consider more popular languages;
C++ has probably 1000 times the number of users... so I may expect roughly
2000 competing solutions? Nope, there's only one (not even 2).

Java, there are 2, but the obvious winner is maintained by Red Hat.
Rust, there are 2, and no clear indicator which to choose (same problem).
Etc.

Anyway, I think it's a bad situation, and 3 colleagues have echoed the same
sentiment independently. I could tolerate if there were meaningful
differences between the solutions, but they're effectively the same
solution, just written+maintained by different individuals.

Projects with multiple contributors are usually stronger than single-person
projects.
Competing one-man projects shooting it out in the wild-west is not
confidence inspiring at all, a single agreed solution supplied by an
authority (ie, "dlang.org") or something is preferred. If some enthusiastic
user wants to develop a meaningfully superior solution, power to them, but
the former should exist for the conservative minds. Most people don't want
to experiment, or take time trying to work out which solution is the better
or accepted standard, they explicitly prefer to conform, and that should be
the easiest path.
Jul 04 2017
prev sibling next sibling parent reply Wulfklaue <wulfklaue wulfklaue.com> writes:
On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:

 I've been using code-d for a while, and it generally works 
 well. I've also noticed there's another plugin available, which 
 at the time I first looked, appeared to be older and 
 less-featured than code-d.


dlang-vscode.dlang seems to be more alive at pure minimum. 
Webfreak his code-d is more up to date and more easy to install. 
Well, months ago it was my impression.

The issue is sometimes some plugin authors take breaks. And 
sometimes the latest versions of D break something, making it 
impossible for the plugins to compile the dependencies. So its 
better the have choice then.

Is it better to have one plugin where both authors work upon. 
Yes. Will it happen: Probably not.
Jul 04 2017
parent Manu via Digitalmars-d <digitalmars-d puremagic.com> writes:
On 4 July 2017 at 23:27, Wulfklaue via Digitalmars-d <
digitalmars-d puremagic.com> wrote:


 On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:


 I've been using code-d for a while, and it generally works well. I've
 also noticed there's another plugin available, which at the time I first
 looked, appeared to be older and less-featured than code-d.


 dlang-vscode.dlang seems to be more alive at pure minimum. Webfreak his
 code-d is more up to date and more easy to install. Well, months ago it was
 my impression.

 The issue is sometimes some plugin authors take breaks. And sometimes the
 latest versions of D break something, making it impossible for the plugins
 to compile the dependencies. So its better the have choice then.

 Is it better to have one plugin where both authors work upon. Yes. Will it
 happen: Probably not.


Right, and this is why I'm trying to make a plea to the respective authors;
please consider merging the projects. The situation is bad. If the only
meaningful factor keeping these things separate is ego, then it's
objectively bad for the community.
Jul 04 2017
prev sibling next sibling parent reply bitwise <bitwise.pvt gmail.com> writes:
On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:

 My feeling is, having 2 very similar plugins is confusing to 
 potential
 users, and it undermines user confidence. Ie, users have the 
 feeling that
 they're competing hacky things maintained by some guy, rather 
 than
 something that's more 'official' with consolidated community 
 support.


I have to disagree with this. When I search for plugins and see a 
wide selection of completing products, it tells me that the 
language has a healthy ecosystem, with a lot of interested users. 
This in turn gives me faith in the language, and it's future, and 
makes me more comfortable investing time in it. I'm much more 
likely to forgive the bugs I find on the grounds that they will 
most likely be fixed.

On the other hand, if there was only one plugin, and it was the 
"official" plugin, branded with "D Language Foundation" or 
"Digital Mars", this would paint a different, yet also positive 
picture for me.
Jul 04 2017
parent reply Manu via Digitalmars-d <digitalmars-d puremagic.com> writes:
On 5 July 2017 at 12:37, bitwise via Digitalmars-d <
digitalmars-d puremagic.com> wrote:


 On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:


 My feeling is, having 2 very similar plugins is confusing to potential
 users, and it undermines user confidence. Ie, users have the feeling that
 they're competing hacky things maintained by some guy, rather than
 something that's more 'official' with consolidated community support.


 I have to disagree with this. When I search for plugins and see a wide
 selection of completing products, it tells me that the language has a
 healthy ecosystem, with a lot of interested users. This in turn gives me
 faith in the language, and it's future, and makes me more comfortable
 investing time in it. I'm much more likely to forgive the bugs I find on
 the grounds that they will most likely be fixed.

 On the other hand, if there was only one plugin, and it was the "official"
 plugin, branded with "D Language Foundation" or "Digital Mars", this would
 paint a different, yet also positive picture for me.


I'm glad we can agree on that second point :)
On the former point; I personally can sympathise with that opinion only in
the context of being a nerd. It is my experience that most professional
programmers I've ever worked with are NOT nerds, they are just people who
have a job. Most people aren't interested in 'ecosystem health', they are
interested in authority, and intend/prefer to conform with the accepted
standards. The worst thing they can see is a 'vibrant' ecosystem. It looks
like 'linux nonsense' to these people.
Jul 04 2017
parent reply MakersF <makers.f.dev gmail.com> writes:
On Wednesday, 5 July 2017 at 02:46:05 UTC, Manu wrote:

 On 5 July 2017 at 12:37, bitwise via Digitalmars-d < 
 digitalmars-d puremagic.com> wrote:


 On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:


 My feeling is, having 2 very similar plugins is confusing to 
 potential users, and it undermines user confidence. Ie, users 
 have the feeling that they're competing hacky things 
 maintained by some guy, rather than something that's more 
 'official' with consolidated community support.


 I have to disagree with this. When I search for plugins and 
 see a wide selection of completing products, it tells me that 
 the language has a healthy ecosystem, with a lot of interested 
 users. This in turn gives me faith in the language, and it's 
 future, and makes me more comfortable investing time in it. 
 I'm much more likely to forgive the bugs I find on the grounds 
 that they will most likely be fixed.

 On the other hand, if there was only one plugin, and it was 
 the "official" plugin, branded with "D Language Foundation" or 
 "Digital Mars", this would paint a different, yet also 
 positive picture for me.


 I'm glad we can agree on that second point :)
 On the former point; I personally can sympathise with that 
 opinion only in
 the context of being a nerd. It is my experience that most 
 professional
 programmers I've ever worked with are NOT nerds, they are just 
 people who
 have a job. Most people aren't interested in 'ecosystem 
 health', they are
 interested in authority, and intend/prefer to conform with the 
 accepted
 standards. The worst thing they can see is a 'vibrant' 
 ecosystem. It looks
 like 'linux nonsense' to these people.


I have to agree with that.


As you can see both plugins are wrappers around other tools 
(good, no wheel reinvention), and they offer some functionality 
on top.
All that follows applies because of this: using the same tools 
underneath means that the same results are provided. There is no 
difference in functionality, only on how it is exposed to the 
user.

https://github.com/Pure-D/code-d/blob/master/README.md

https://github.com/dlang-vscode/dlang-vscode/blob/master/README.md

The main difference (and I'm fairly sure pain point as well for 
new users, at least for me) is how these tools are installed or 
expected on the system.

UX is something that takes a bit of time and it's not always fun 
(need to test many different configurations, support different 
systems which do things in different ways) and I really see no 
point in duplicating this tedious work.

And if you look at the contributors tab, you see that both 
projects are not in a healthy state: one is developed solely by 
WebFreak001 and the other is split 3/4 and 1/4 between the two 
authors.
Which means that if one of them leaves the project (because of 
work,family, persona issues, boredom, any situation which might 
happen in life) the project is likely going to stagnate and 
become dead.

Having instead 3 active people on a project means that a lot of 
improvement and consistent work can be put into it, also easing 
the work needed to accept PR from the community.

I understand and appreciate the amount of work the developers of 
the two extensions put into them, but I think it's fair to accept 
that the current situation is not the best one for the users and 
the community.
Jul 05 2017
parent reply bitwise <bitwise.pvt gmail.com> writes:
On Wednesday, 5 July 2017 at 08:18:59 UTC, MakersF wrote:

 Having instead 3 active people on a project means that a lot of 
 improvement and consistent work can be put into it, also easing 
 the work needed to accept PR from the community.


Which is why it would be great to see 3+ active developers on 
each one ;)
Jul 05 2017
parent Wulfklaue <wulfklaue wulfklaue.com> writes:
On Wednesday, 5 July 2017 at 18:12:21 UTC, bitwise wrote:

 Which is why it would be great to see 3+ active developers on 
 each one ;)


Take in account that WebFreak001 is working on serve-d...


 Microsoft language server protocol implementation for D using 
 workspace-d.
 This program is basically a combination of workspace-d and most 
 of code-d.
 The purpose of this project is to give every editor the same 
 capabilities and editing  features as code-d with even less 
 code required on the editor side than with workspace-d due to a 
 more widely available protocol.
 This is pretty much another abstraction layer on top of 
 workspace-d to simplify and speed up extension development as 
 most of the editor extension can now be written in D.


His idea sound like a better future design...
Jul 05 2017
prev sibling next sibling parent Moritz Maxeiner <moritz ucworks.org> writes:
On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:

 So, is there a reason not to merge the projects beyond ego?


While both share the same tools used by pretty much all 
editor/IDE plugins for D, dlang-vscode requires *only* those, 
while code-d additionally requires workspace-d.
YMMV, but that is why I used dlang-vscode exclusively.
Jul 05 2017
prev sibling parent reply WebFreak001 <d.forum webfreak.org> writes:
On Tuesday, 4 July 2017 at 04:49:29 UTC, Manu wrote:

 I've been using code-d for a while, and it generally works 
 well. I've also noticed there's another plugin available, which 
 at the time I first looked, appeared to be older and 
 less-featured than code-d.

 I've recently had a couple of colleagues ask me which plugin to 
 install, and I noticed that both seem to be up-to-date these 
 days, and this leads to confusion.

 Looking at the feature list, it appears that both plugins do 
 mostly the
 same stuff.
 My feeling is, having 2 very similar plugins is confusing to 
 potential
 users, and it undermines user confidence. Ie, users have the 
 feeling that
 they're competing hacky things maintained by some guy, rather 
 than
 something that's more 'official' with consolidated community 
 support. I
 also tend to presume in these situations that the 'proper' one 
 is the one
 with the most users/installs, but that's not clear either in 
 this case.
 I know this has nothing to do with the truth, but it's about 
 perception and
 first impressions. Little things matter.

 If authors of both plugins are active here, I ask; why have 2 
 separate
 plugins?
 I can't imagine any reason for divergence. I would be a lot more
 comfortable if there was only one with multiple contributors. 
 Projects with
 many contributors always inspire a lot more confidence than 
 multiple
 overlapping projects with one contributor each...

 So, is there a reason not to merge the projects beyond ego?


hi, code-d author here.

I personally think a merge of all of our features into one plugin 
would be a great idea but the problem is that the plugins work a 
lot differently in the code level. My plugin uses workspace-d to 
abstract dfmt, dcd and dscanner so the extension doesn't need to 
know about it and dlang-vscode simply uses the extensions 
directly. I am also planning on further abstracting code-d using 
serve-d and Microsoft's Language Server Protocol for a wider 
adoption and better support across editors, but that also means 
that the extension code will be mostly thrown out and converted 
to D. While this isn't such a big problem for me because most 
code was already in workspace-d, I think if we implemented all 
the features from dlang-vscode it would be a lot of work not to 
break things. And doing it the other way around (merging my code 
into dlang-vscode) would mean that all the abstraction goes away 
and that it would be a vscode only plugin. code-d is basically 
the same as the one for atom or sublime text, just with all the 
features implemented instead of just a subset. While I didn't 
really support atom or sublime text for a while, I still think 
it's a good idea to keep these projects around for anyone who 
wants to implement workspace-d or serve-d into their extension as 
example reference how to implement them.

But if the authors of dlang-vscode could contact me on IRC for 
example (WebFreak on #d on freenode) we might be able to find a 
solution to improve the situation and either merge into one 
plugin or share some code and ideas.
Jul 06 2017
parent Jerry <hurricane hereiam.com> writes:
On Thursday, 6 July 2017 at 16:23:06 UTC, WebFreak001 wrote:

 While this isn't such a big problem for me because most code 
 was already in workspace-d,


How much is actually there though? Going by lines of code, 
dlang-vscode is a total of ~2000 lines of code *less* than 
code-d. That's not including workspace-d lines of code for code-d 
as well. So it can provide the same features in far fewer lines 
and with a less complex system. Having an extra process just 
isn't worth it the abstraction you get cause it is so minimal. I 
wouldn't be trying to abstract it even further, you are just over 
complicating something for no reason. Keep it simple.
Jul 14 2017