www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Use proper frameworks for building dlang.org

reply Jacob Carlborg <doob me.com> writes:
Lately Andrei has worked a lot with improving the dlang.org site in 
various ways. To me it getting more clear and clear that Ddoc is not the 
right tool for building a web site. Especially the latest "improvement" 
[1] shows that it's not a good idea to reinvent the wheel, especially 
when it's not an improvement, at all.

Why don't we instead make use of a proper framework both on the server 
side and client side. Personally I would go with Ruby on Rails but I 
know that most of you here would hate that so a better suggestion would 
probably be vibe.d. For the client side I'm thinking Bootstrap and jQuery.

The biggest reason why I would prefer Rails is because I know everything 
that is needed is already implemented and easily available. I can not 
say the same thing about vibe.d. But it might be enough for dlang.org, I 
don't know.

What do you think?

[1] http://forum.dlang.org/thread/m9f558$lbb$1 digitalmars.com

-- 
/Jacob Carlborg
Jan 18 2015
next sibling parent "aldanor" <i.s.smirnov gmail.com> writes:
On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg wrote:
 Lately Andrei has worked a lot with improving the dlang.org 
 site in various ways. To me it getting more clear and clear 
 that Ddoc is not the right tool for building a web site. 
 Especially the latest "improvement" [1] shows that it's not a 
 good idea to reinvent the wheel, especially when it's not an 
 improvement, at all.

 Why don't we instead make use of a proper framework both on the 
 server side and client side. Personally I would go with Ruby on 
 Rails but I know that most of you here would hate that so a 
 better suggestion would probably be vibe.d. For the client side 
 I'm thinking Bootstrap and jQuery.

 The biggest reason why I would prefer Rails is because I know 
 everything that is needed is already implemented and easily 
 available. I can not say the same thing about vibe.d. But it 
 might be enough for dlang.org, I don't know.

 What do you think?

 [1] http://forum.dlang.org/thread/m9f558$lbb$1 digitalmars.com
I would agree on this in general. Not sure about Rails (because it is totally an overkill and may be a pain to maintain) but everyone has their own preferences for the backend and templating engine.... in this case even a proper static generator (Jekyll if you're talking Ruby) with a proper responsive front end (JQuery + Bootstrap / Foundation / Semantic UI / whichever) would do. If it's markdown-based, it would also significantly lower the barrier for people other than a few devs who would like to contribute to the docs and website but would hate to do it in ddoc.
Jan 18 2015
prev sibling next sibling parent "Sebastiaan Koppe" <mail skoppe.eu> writes:
On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg wrote:
 Lately Andrei has worked a lot with improving the dlang.org 
 site in various ways. To me it getting more clear and clear 
 that Ddoc is not the right tool for building a web site. 
 Especially the latest "improvement" [1] shows that it's not a 
 good idea to reinvent the wheel, especially when it's not an 
 improvement, at all.

 Why don't we instead make use of a proper framework both on the 
 server side and client side. Personally I would go with Ruby on 
 Rails but I know that most of you here would hate that so a 
 better suggestion would probably be vibe.d. For the client side 
 I'm thinking Bootstrap and jQuery.

 The biggest reason why I would prefer Rails is because I know 
 everything that is needed is already implemented and easily 
 available. I can not say the same thing about vibe.d. But it 
 might be enough for dlang.org, I don't know.

 What do you think?

 [1] http://forum.dlang.org/thread/m9f558$lbb$1 digitalmars.com
Vibe.d is already used to pre-render the site from ddox to html as it is. It's usage can probably be extended to serving the whole site. It is a good idea since any tooling you might build for this site, can then be reused by other people.
Jan 18 2015
prev sibling next sibling parent reply "MattCoder" <stop spam.com> writes:
On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg wrote:
 What do you think?
I'm seeing a lot a topics regards about fixing website, styles an so on. Maybe is time to try to raise money and hire someone with good knowledge and of course freetime to work only on it? Matheus.
Jan 18 2015
parent reply "aldanor" <i.s.smirnov gmail.com> writes:
On Sunday, 18 January 2015 at 13:01:48 UTC, MattCoder wrote:
 On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg 
 wrote:
 What do you think?
I'm seeing a lot a topics regards about fixing website, styles an so on. Maybe is time to try to raise money and hire someone with good knowledge and of course freetime to work only on it? Matheus.
It also looks a bit wrong seeing Andrei as one of the core devs putting so much time into the website issues, styling in particular (don't get me wrong, this is absolutely fantastic that someone's willing to do that, thanks Andrei -- and it looks like most of the community now recognizes the need). It's that any semi-decent web designer would probably do a better job on the website design and layout while still spending less time doing that (because they've done it a hundred times already and/or do it for living). However, not every web designer can hack on system language internals and review D PRs!.. So someone needs to step up and do it (and/or redo it, and/or hire someone to do it). That's just my 2 cents which is purely subjective.
Jan 18 2015
next sibling parent reply "weaselcat" <weaselcat gmail.com> writes:
On Sunday, 18 January 2015 at 14:00:57 UTC, aldanor wrote:
 On Sunday, 18 January 2015 at 13:01:48 UTC, MattCoder wrote:
 On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg 
 wrote:
 What do you think?
I'm seeing a lot a topics regards about fixing website, styles an so on. Maybe is time to try to raise money and hire someone with good knowledge and of course freetime to work only on it? Matheus.
It also looks a bit wrong seeing Andrei as one of the core devs putting so much time into the website issues, styling in particular (don't get me wrong, this is absolutely fantastic that someone's willing to do that, thanks Andrei -- and it looks like most of the community now recognizes the need). It's that any semi-decent web designer would probably do a better job on the website design and layout while still spending less time doing that (because they've done it a hundred times already and/or do it for living). However, not every web designer can hack on system language internals and review D PRs!.. So someone needs to step up and do it (and/or redo it, and/or hire someone to do it). That's just my 2 cents which is purely subjective.
Nobody else has really stepped up to do it, so I think Andrei just took it upon himself to do it. If there was a crowdfunding to pay for designers for a new D website I'd definitely chip in a bit, but I don't know how far that would go. I agree that the current site seems like an ugly ddoc hack, I just recently tried to fix a broken link on the frontpage(should be easy, right?) and was confused by ddoc so I had to submit a bug request.
Jan 18 2015
next sibling parent reply "MattCoder" <stop spam.com> writes:
On Sunday, 18 January 2015 at 14:06:06 UTC, weaselcat wrote:
 ... I agree that the current site seems like an ugly ddoc hack, 
 I just recently tried to fix a broken link on the 
 frontpage(should be easy, right?) and was confused by ddoc so I 
 had to submit a bug request.
You know, I quit my job for my own wish, you know I want a fresh air and do others things this year. I could try to help with the site, but there is two main problems: 1) I don't have that UX skills like those professionals out there. 2) I don't know where to start, and I'd like to ask if there is a doc anywhere teaching how to contribute, for example this site, if I want to try some experiments where should I start? Matheus.
Jan 18 2015
parent "MattCoder" <stop spam.com> writes:
On Sunday, 18 January 2015 at 15:27:36 UTC, MattCoder wrote:
 You know, I quit my job for my own wish, you know I want a 
 fresh air and do others things this year. I could try to help 
 with the site, but there is two main problems:
Sorry I'm not english, so I think the text above isn't quite right, so: "You know, I quit my job for own will, you know, I was looking for a fresh air and do others things this year. I could try to help with the site, but there are two main problems:" Matheus.
Jan 18 2015
prev sibling parent Andrei Alexandrescu <SeeWebsiteForEmail erdan.org> writes:
"weaselcat" <weaselcat gmail.com> wrote:
 On Sunday, 18 January 2015 at 14:00:57 UTC, aldanor wrote:
 On Sunday, 18 January 2015 at 13:01:48 UTC, MattCoder wrote:
 On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg >> wrote:
 What do you think?
I'm seeing a lot a topics regards about fixing website, styles >> an so on. Maybe is time to try to raise money and hire someone >> with good knowledge and of course freetime to work only on it? Matheus.
It also looks a bit wrong seeing Andrei as one of the core devs > putting so much time into the website issues, styling in > particular (don't get me wrong, this is absolutely fantastic > that someone's willing to do that, thanks Andrei -- and it > looks like most of the community now recognizes the need). It's that any semi-decent web designer would probably do a > better job on the website design and layout while still > spending less time doing that (because they've done it a > hundred times already and/or do it for living). However, not > every web designer can hack on system language internals and > review D PRs!.. So someone needs to step up and do it (and/or > redo it, and/or hire someone to do it). That's just my 2 cents which is purely subjective.
Nobody else has really stepped up to do it, so I think Andrei just took it upon himself to do it. If there was a crowdfunding to pay for designers for a new D website I'd definitely chip in a bit, but I don't know how far that would go. I agree that the current site seems like an ugly ddoc hack, I just recently tried to fix a broken link on the frontpage(should be easy, right?) and was confused by ddoc so I had to submit a bug request.
The main problem with the site right now is the content. No amount of frameworks will improve that. Many pages are just old and obsolete. I'll look into rewriting some docs. In the meantime styling would be nice to improve. I wonder what are the typical costs for such work. Thoughts? Where is the bug? Changing a macro is trivial.
Jan 18 2015
prev sibling next sibling parent "MattCoder" <stop spam.com> writes:
On Sunday, 18 January 2015 at 14:00:57 UTC, aldanor wrote:
 On Sunday, 18 January 2015 at 13:01:48 UTC, MattCoder wrote:
 On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg 
 wrote:
 What do you think?
I'm seeing a lot a topics regards about fixing website, styles an so on. Maybe is time to try to raise money and hire someone with good knowledge and of course freetime to work only on it? Matheus.
It also looks a bit wrong seeing Andrei as one of the core devs putting so much time into the website issues, styling in particular (don't get me wrong, this is absolutely fantastic that someone's willing to do that, thanks Andrei -- and it looks like most of the community now recognizes the need).
Sure and that are my feelings too. And of course we all recognize Andrei's efforts. Matheus.
Jan 18 2015
prev sibling parent Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/18/15 6:00 AM, aldanor wrote:
 On Sunday, 18 January 2015 at 13:01:48 UTC, MattCoder wrote:
 On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg wrote:
 What do you think?
I'm seeing a lot a topics regards about fixing website, styles an so on. Maybe is time to try to raise money and hire someone with good knowledge and of course freetime to work only on it? Matheus.
It also looks a bit wrong seeing Andrei as one of the core devs putting so much time into the website issues, styling in particular (don't get me wrong, this is absolutely fantastic that someone's willing to do that, thanks Andrei -- and it looks like most of the community now recognizes the need).
To be frank, that's more than a bit wrong. Let me put this in perspective. We're facing a momentous year with ambitious goals, stiff competition, and a growing community to support. At the beginning of January I wanted to create a "vision" document, publish it here, and talk about it this Thursday at http://www.meetup.com/D-Lang-Sillicon-Valley/events/219413448/ (be there!). That kind of work requires a certain state of mind and some good getting into it, and it's already past mid January with nothing to show. Also, I am looking at creating a D Language Foundation. A lawyer is waiting for information from me since January 5. Not to mention a bunch of work on Phobos (relational algebra, container stuff, reviewing work that's waiting to be reviewed, documentation again). I also wanted to write "The Range Bible" that formalizes everything there is about ranges. Hefty goals. All of that didn't happen because I got into fixing things on the website. It started with fixes for a few blatant issues and breakages, and it only got more involved. As I'm sure it's known to all of us that kind of stuff gets you into a flow that's damaging to break. I figured if I let myself interrupted it'll take months before I get back to it and nobody else will, so I pushed through a number of things until the menus of yesterday. My plan is to "outsource" as much of the site as possible to CSS, then ask a designer to improve the CSS. People can do pretty amazing things with those. After the recent work I did we have almost no "raw" HTML tags in the site docs - all are either styled with classes or semantically meaningful. Anyhow, quite literally I don't have the time for vision because I'm too busy fixing the website. The perhaps less productive factor is the constant bickering on the newsgroup. For every line of code it seems there's 30 people to comment on it, but too few actually write a better one. Shoveling crap once in a while is vigorous exercise, but doing it surrounded by sage people sitting on their testes and telling how it should be done is... quite fun :o).
 It's that any semi-decent web designer would probably do a better job on
 the website design and layout while still spending less time doing that
 (because they've done it a hundred times already and/or do it for
 living). However, not every web designer can hack on system language
 internals and review D PRs!.. So someone needs to step up and do it
 (and/or redo it, and/or hire someone to do it).

 That's just my 2 cents which is purely subjective.
That's my plan exactly (I'll talk more on this). Problem is, until recently the site was not in shape to be restyled because it used a bunch of ad-hoc inconsistent styling. Right now a lot more is factored into the CSS, and it should be a modular job to ask a professional designer to redo the CSS. Andrei
Jan 18 2015
prev sibling next sibling parent reply "Mathias LANG" <geod24 gmail.com> writes:
On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg wrote:
 Lately Andrei has worked a lot with improving the dlang.org 
 site in various ways. To me it getting more clear and clear 
 that Ddoc is not the right tool for building a web site. 
 Especially the latest "improvement" [1] shows that it's not a 
 good idea to reinvent the wheel, especially when it's not an 
 improvement, at all.

 Why don't we instead make use of a proper framework both on the 
 server side and client side. Personally I would go with Ruby on 
 Rails but I know that most of you here would hate that so a 
 better suggestion would probably be vibe.d. For the client side 
 I'm thinking Bootstrap and jQuery.

 The biggest reason why I would prefer Rails is because I know 
 everything that is needed is already implemented and easily 
 available. I can not say the same thing about vibe.d. But it 
 might be enough for dlang.org, I don't know.

 What do you think?

 [1] http://forum.dlang.org/thread/m9f558$lbb$1 digitalmars.com
I started to experiment with this. The problem we'll get is that DDOC is not used only for the website. It generates kindle / mobi / pdf. So I see 3 paths here: 1) Implement support for generating the existing target using Vibe.d. 2) Use DDOC pages as the content, and make Vibe.d use that. 3) Get rid of the other targets, and distribute Ali's book to those who want offline references. IMO 1) is way too much work and too few benefit, 2) seems brittle, so I would go with 3. But we would have to get Andrei and Walter's approval.
Jan 18 2015
next sibling parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/18/15 6:32 AM, Mathias LANG wrote:
 On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg wrote:
 Lately Andrei has worked a lot with improving the dlang.org site in
 various ways. To me it getting more clear and clear that Ddoc is not
 the right tool for building a web site. Especially the latest
 "improvement" [1] shows that it's not a good idea to reinvent the
 wheel, especially when it's not an improvement, at all.

 Why don't we instead make use of a proper framework both on the server
 side and client side. Personally I would go with Ruby on Rails but I
 know that most of you here would hate that so a better suggestion
 would probably be vibe.d. For the client side I'm thinking Bootstrap
 and jQuery.

 The biggest reason why I would prefer Rails is because I know
 everything that is needed is already implemented and easily available.
 I can not say the same thing about vibe.d. But it might be enough for
 dlang.org, I don't know.

 What do you think?

 [1] http://forum.dlang.org/thread/m9f558$lbb$1 digitalmars.com
I started to experiment with this. The problem we'll get is that DDOC is not used only for the website. It generates kindle / mobi / pdf. So I see 3 paths here: 1) Implement support for generating the existing target using Vibe.d. 2) Use DDOC pages as the content, and make Vibe.d use that. 3) Get rid of the other targets, and distribute Ali's book to those who want offline references. IMO 1) is way too much work and too few benefit, 2) seems brittle, so I would go with 3. But we would have to get Andrei and Walter's approval.
I don't know. So what would be the replacement? BTW turns out that dlangspec.pdf is downloaded quite a bit. -- Andrei
Jan 18 2015
parent reply "Mathias LANG" <geod24 gmail.com> writes:
On Sunday, 18 January 2015 at 16:41:36 UTC, Andrei Alexandrescu 
wrote:
 [...]
 I don't know. So what would be the replacement? BTW turns out 
 that dlangspec.pdf is downloaded quite a bit. -- Andrei
Bummer. I never used it, so I figured I should give it a try before attempting to kill it. I didn't see it at first sight on the download page (which is quite inconsistent), but it's there. So I simply searched the website. First result was a link to your post on the newsgroup, second was this: http://dlang.org/dlangspec.pdf No link to downloads or specs on the first page of google results (but that might be different for others). ATM, it's 4 completely useless pages, and seem to have been this way for quite some time: http://forum.dlang.org/thread/yqhwwpskwmkdefarjavm forum.dlang.org Searching Github yield no confidence in it either ( https://github.com/D-Programming-Language/dlang.org/pull/687 ) How many maintainers actively work on the pdf / mobi version ? From what I can grep from dlang.org's commit, you are the only one taking care of this. I'm just saying a pdf/mobi version is not important, or there's no interest in it. All I'm saying is that Ali poured years of work to write and keep his book up to date. So why don't we use that ?
Jan 19 2015
parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/19/15 2:49 AM, Mathias LANG wrote:
 On Sunday, 18 January 2015 at 16:41:36 UTC, Andrei Alexandrescu wrote:
 [...]
 I don't know. So what would be the replacement? BTW turns out that
 dlangspec.pdf is downloaded quite a bit. -- Andrei
Bummer. I never used it, so I figured I should give it a try before attempting to kill it. I didn't see it at first sight on the download page (which is quite inconsistent), but it's there. So I simply searched the website. First result was a link to your post on the newsgroup, second was this: http://dlang.org/dlangspec.pdf No link to downloads or specs on the first page of google results (but that might be different for others). ATM, it's 4 completely useless pages, and seem to have been this way for quite some time: http://forum.dlang.org/thread/yqhwwpskwmkdefarjavm forum.dlang.org Searching Github yield no confidence in it either ( https://github.com/D-Programming-Language/dlang.org/pull/687 )
That is recent. Many changes to content break the pdf build and I didn't have the time to fix it.
 How many maintainers actively work on the pdf / mobi version ? From what
 I can grep from dlang.org's commit, you are the only one taking care of
 this.

 I'm just saying a pdf/mobi version is not important, or there's no
 interest in it.
 All I'm saying is that Ali poured years of work to write and keep his
 book up to date. So why don't we use that ?
You need both tutorial and reference. But that's beside the point. This seems to suggest the choice is "if we didn't have to build the mobi and pdf, the road is clear to using Great Framework(tm)". There are a few issues with that, such as nobody seems to agree what the Great Framework is, though I'd happily switch to vibe.d for obvious reasons if there were a champion for it; second, after all is said and done it's me to pay the piper. Andrei
Jan 19 2015
parent reply "Mathias LANG" <geod24 gmail.com> writes:
On Monday, 19 January 2015 at 16:38:58 UTC, Andrei Alexandrescu 
wrote:
 But that's beside the point. This seems to suggest the choice 
 is "if we didn't have to build the mobi and pdf, the road is 
 clear to using Great Framework(tm)". There are a few issues 
 with that, such as nobody seems to agree what the Great 
 Framework is, though I'd happily switch to vibe.d for obvious 
 reasons if there were a champion for it; second, after all is 
 said and done it's me to pay the piper.


 Andrei
The reason I'm asking is because this is exactly what I started working on. While not a web designer, I'm sure I have enough experience with Vibe to get this done. However, I just wanted to make sure it has a chance to go in. Glad to hear it is ! I'll work on a PoC and let you know. I'll try to do it ASAP, however I didn't want to make any promise, as I'm finishing my master and will start a job abroad in March, so it's quite hard to plan things. Anyway, for any suggestion / concern / ping, my email is easily found here or on Github ( Geod24).
Jan 19 2015
parent Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/19/15 8:57 AM, Mathias LANG wrote:
 On Monday, 19 January 2015 at 16:38:58 UTC, Andrei Alexandrescu wrote:
 But that's beside the point. This seems to suggest the choice is "if
 we didn't have to build the mobi and pdf, the road is clear to using
 Great Framework(tm)". There are a few issues with that, such as nobody
 seems to agree what the Great Framework is, though I'd happily switch
 to vibe.d for obvious reasons if there were a champion for it; second,
 after all is said and done it's me to pay the piper.


 Andrei
The reason I'm asking is because this is exactly what I started working on. While not a web designer, I'm sure I have enough experience with Vibe to get this done. However, I just wanted to make sure it has a chance to go in. Glad to hear it is !
By all means. Thanks!!
 I'll work on a PoC and let you know. I'll try to do it ASAP, however I
 didn't want to make any promise, as I'm finishing my master and will
 start a job abroad in March, so it's quite hard to plan things.
 Anyway, for any suggestion / concern / ping, my email is easily found
 here or on Github ( Geod24).
Sounds good. I think a good transition path will be crucial. Andrei
Jan 19 2015
prev sibling parent Jacob Carlborg <doob me.com> writes:
On 2015-01-18 15:32, Mathias LANG wrote:

 I started to experiment with this. The problem we'll get is that DDOC is
 not used only for the website. It generates kindle / mobi / pdf.
 So I see 3 paths here:
 1) Implement support for generating the existing target using Vibe.d.
 2) Use DDOC pages as the content, and make Vibe.d use that.
 3) Get rid of the other targets, and distribute Ali's book to those who
 want offline references.

 IMO 1) is way too much work and too few benefit, 2) seems brittle, so I
 would go with 3. But we would have to get Andrei and Walter's approval.
I would go with 3). What's the reason to have a website in kindle / mobi / pdf? If needed, can't that be generated from the HTML? -- /Jacob Carlborg
Jan 18 2015
prev sibling next sibling parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/18/15 2:24 AM, Jacob Carlborg wrote:
 Lately Andrei has worked a lot with improving the dlang.org site in
 various ways. To me it getting more clear and clear that Ddoc is not the
 right tool for building a web site. Especially the latest "improvement"
 [1] shows that it's not a good idea to reinvent the wheel, especially
 when it's not an improvement, at all.
Heheh, way to use Archuduke's Ferdinand assassination :o). I'm sorry Jacob but this one is really out there. DDoc has nothing with this, and to the extent it does it actually helps. So I start yesterday with searching "Css vertical menus" and the such. After some browsing I figured a sort of hierarchical/accordion menu would be good for us. There's plenty of them to choose from, neither of which require Rails - apparently there are many lost souls out there. I insisted on one that doesn't need Javascript, but at that point I couldn't find any. So I settled for those that use jQuery, which we already use so - nice. There are plenty of those too. So I chose one, downloaded, and integrated it within our site. The funny part is it was easier to work with the thing _because_ of ddoc, not in spite of it. So our source code for menus looks like this: http://goo.gl/QddFkh. It makes it very easy to move entries around, change them all at once, etc. That generates http://goo.gl/ZLJisi, which even after nice formatting is quite difficult to work with. I might be ignorant but I have difficulty figuring exactly where ddoc's fault is. I literally had a big chuckle when I saw you attempting to use my poor taste in colors as a reason to ditch ddoc.
 Why don't we instead make use of a proper framework both on the server
 side and client side. Personally I would go with Ruby on Rails but I
 know that most of you here would hate that so a better suggestion would
 probably be vibe.d. For the client side I'm thinking Bootstrap and jQuery.
Wait, jQuery is the pitts! No?
 The biggest reason why I would prefer Rails is because I know everything
 that is needed is already implemented and easily available. I can not
 say the same thing about vibe.d. But it might be enough for dlang.org, I
 don't know.

 What do you think?

 [1] http://forum.dlang.org/thread/m9f558$lbb$1 digitalmars.com
If we change the tooling, I'd say vibe.d is the best option for obvious dogfooding reasons. The problem with Rails and other frameworks is that *you* prefer it, but *I* and *others* are in for most of the work with it. Andrei
Jan 18 2015
parent reply Jacob Carlborg <doob me.com> writes:
On 2015-01-18 18:23, Andrei Alexandrescu wrote:

 Heheh, way to use Archuduke's Ferdinand assassination :o). I'm sorry
 Jacob but this one is really out there.

 DDoc has nothing with this, and to the extent it does it actually helps.
No it has not, but other strange solutions you're coming up with for CSS minification and gzip. I'm not going to touch the code for the web site as long as it's written in Ddoc. Simple as that.
 So I start yesterday with searching "Css vertical menus" and the such.
 After some browsing I figured a sort of hierarchical/accordion menu
 would be good for us. There's plenty of them to choose from, neither of
 which require Rails - apparently there are many lost souls out there. I
 insisted on one that doesn't need Javascript, but at that point I
 couldn't find any. So I settled for those that use jQuery, which we
 already use so - nice. There are plenty of those too.

 So I chose one, downloaded, and integrated it within our site. The funny
 part is it was easier to work with the thing _because_ of ddoc, not in
 spite of it. So our source code for menus looks like this:
 http://goo.gl/QddFkh. It makes it very easy to move entries around,
 change them all at once, etc. That generates http://goo.gl/ZLJisi, which
 even after nice formatting is quite difficult to work with.
I don't understand this? Do you think the alternative of Ddoc is to use plain HTML? I would never use just plain HMTL. I would use a server side language to generate it. With all the flexibility it comes with it.
 If we change the tooling, I'd say vibe.d is the best option for obvious
 dogfooding reasons. The problem with Rails and other frameworks is that
 *you* prefer it, but *I* and *others* are in for most of the work with it.
That's why I said vibe.d would be a better choice. -- /Jacob Carlborg
Jan 18 2015
next sibling parent "Adam D. Ruppe" <destructionator gmail.com> writes:
On Sunday, 18 January 2015 at 18:49:43 UTC, Jacob Carlborg wrote:
 No it has not, but other strange solutions you're coming up 
 with for CSS minification and gzip.
That's just a server config issue and can be fixed with 15 lines of .htaccess and like three lines of shell. I posted a solution last night, and the only reason I didn't implement it on the live server is I don't have access to it and don't know the procedure to get stuff up there through a middle man...
Jan 18 2015
prev sibling parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/18/15 10:49 AM, Jacob Carlborg wrote:
 On 2015-01-18 18:23, Andrei Alexandrescu wrote:

 Heheh, way to use Archuduke's Ferdinand assassination :o). I'm sorry
 Jacob but this one is really out there.

 DDoc has nothing with this, and to the extent it does it actually helps.
No it has not, but other strange solutions you're coming up with for CSS minification and gzip.
So now minification and gzipping are the culprit? I don't quite understand.
 I'm not going to touch the code for the web site as long as it's written
 in Ddoc. Simple as that.
I understand. Now there are a couple of issues here worth mentioning. 1. Technical arguments are good, but this sounds a bit like an ideological position. I'd have difficulty resonating with that. 2. We all know from math that negating the hypothesis doesn't necessarily lead to negating the conclusion :o). So if we switch to Rails there's no guarantee you will contribute after all; or it's possible you will, just not a lot, and then it's me holding the bag. 3. This rigidity doesn't quite bode well. Simply put, if you led a project, would you want to pave the way to work with a guy with a similar stance? Andrei
Jan 18 2015
parent reply Jacob Carlborg <doob me.com> writes:
On 2015-01-19 03:31, Andrei Alexandrescu wrote:

 So now minification and gzipping are the culprit? I don't quite understand.
There are plenty existing web frameworks that already have solved, what it seems like, all the problems you're trying to solve now. You're just shooting them down because it's not Ddoc.
 I'm not going to touch the code for the web site as long as it's written
 in Ddoc. Simple as that.
I understand. Now there are a couple of issues here worth mentioning. 1. Technical arguments are good, but this sounds a bit like an ideological position. I'd have difficulty resonating with that.
No, the reason is that it's not the right tool for the job. I'm speaking purely technically.
 2. We all know from math that negating the hypothesis doesn't
 necessarily lead to negating the conclusion :o). So if we switch to
 Rails there's no guarantee you will contribute after all; or it's
 possible you will, just not a lot, and then it's me holding the bag.

 3. This rigidity doesn't quite bode well. Simply put, if you led a
 project, would you want to pave the way to work with a guy with a
 similar stance?
As I've already said several times, also form the beginning, that I think vibe.d would be a better choice. -- /Jacob Carlborg
Jan 19 2015
parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/19/15 12:33 AM, Jacob Carlborg wrote:
 On 2015-01-19 03:31, Andrei Alexandrescu wrote:

 So now minification and gzipping are the culprit? I don't quite
 understand.
There are plenty existing web frameworks that already have solved, what it seems like, all the problems you're trying to solve now. You're just shooting them down because it's not Ddoc.
There might be a misunderstanding over what problems I'm trying to solve. -- Andrei
Jan 19 2015
parent reply Ary Borenszweig <ary esperanto.org.ar> writes:
On 1/19/15 1:42 PM, Andrei Alexandrescu wrote:
 On 1/19/15 12:33 AM, Jacob Carlborg wrote:
 On 2015-01-19 03:31, Andrei Alexandrescu wrote:

 So now minification and gzipping are the culprit? I don't quite
 understand.
There are plenty existing web frameworks that already have solved, what it seems like, all the problems you're trying to solve now. You're just shooting them down because it's not Ddoc.
There might be a misunderstanding over what problems I'm trying to solve. -- Andrei
What are you trying to solve?
Jan 19 2015
parent Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/19/15 9:55 AM, Ary Borenszweig wrote:
 On 1/19/15 1:42 PM, Andrei Alexandrescu wrote:
 On 1/19/15 12:33 AM, Jacob Carlborg wrote:
 On 2015-01-19 03:31, Andrei Alexandrescu wrote:

 So now minification and gzipping are the culprit? I don't quite
 understand.
There are plenty existing web frameworks that already have solved, what it seems like, all the problems you're trying to solve now. You're just shooting them down because it's not Ddoc.
There might be a misunderstanding over what problems I'm trying to solve. -- Andrei
What are you trying to solve?
I replied to this once, and forum.dlang.org is down so I'll paste: 1. The site design/styling is old. Simply changing it is now necessary just as a refresher. 2. Of course, we'd also like things to be better. The chaotic navigation menu on the left was something I long wished to replace with a more modern on. 3. Part of my effort is to replace all manual html ad-hoc styles with styles in the css. That is invisible right now but will become very useful when we pass the site for improvement to a specialist. Andrei
Jan 19 2015
prev sibling next sibling parent Ary Borenszweig <ary esperanto.org.ar> writes:
On 1/18/15 7:24 AM, Jacob Carlborg wrote:
 Lately Andrei has worked a lot with improving the dlang.org site in
 various ways. To me it getting more clear and clear that Ddoc is not the
 right tool for building a web site. Especially the latest "improvement"
 [1] shows that it's not a good idea to reinvent the wheel, especially
 when it's not an improvement, at all.

 Why don't we instead make use of a proper framework both on the server
 side and client side. Personally I would go with Ruby on Rails but I
 know that most of you here would hate that so a better suggestion would
 probably be vibe.d. For the client side I'm thinking Bootstrap and jQuery.

 The biggest reason why I would prefer Rails is because I know everything
 that is needed is already implemented and easily available. I can not
 say the same thing about vibe.d. But it might be enough for dlang.org, I
 don't know.

 What do you think?

 [1] http://forum.dlang.org/thread/m9f558$lbb$1 digitalmars.com
I agree with you, of course. But more than anything, you need a real designer.
Jan 18 2015
prev sibling next sibling parent reply "Ola Fosheim =?UTF-8?B?R3LDuHN0YWQi?= writes:
On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg wrote:
 Why don't we instead make use of a proper framework both on the 
 server side and client side. Personally I would go with Ruby on 
 Rails but I know that most of you here would hate that so a 
 better suggestion would probably be vibe.d. For the client side 
 I'm thinking Bootstrap and jQuery.
I would suggest you avoid frameworks since they go out of fashion fairly quickly and makes maintenance dependent on individuals (with framework knowledge). jQuery adds little value since browsers are fairly standards-compliant these days, IMO. http://caniuse.com/#feature_sort=score It would also look very bad if you cannot run dlang.org on D tech. I suggest using dlang.org for driving phobos implementation/binding of standard w3 web tech. No point in having D marketing other languages or their frameworks. 1. semantic markup i XML (DDoc 2 XML would be a good startingpoint) 2. XSLT for transforming XML to HTML5 3. K.I.S.S. design wise: WAI+HTML5+CSS3 (marginal use of JS) Let HTML5 layout design be a group effort and then let 1-2 individuals (who know what they are doing) do the CSS styling from scratch after you have the HTML5 ready.
 The biggest reason why I would prefer Rails is because I know 
 everything that is needed is already implemented and easily 
 available. I can not say the same thing about vibe.d. But it
Yes, but then you+somone else have to maintain it. More people have a interest in learning the basic web standards than learning a specific framework. It would be better to use dlang.org development to drive the design of vibe.d and phobos and front D as a capable server tech. It could also be used for a vibe.d tutorial. A "dlang.org design and build tutorial" could both market D and make it easier for people to contribute.
Jan 19 2015
next sibling parent reply "Sebastiaan Koppe" <mail skoppe.eu> writes:
On Monday, 19 January 2015 at 12:00:58 UTC, Ola Fosheim Grøstad 
wrote:
 On Sunday, 18 January 2015 at 10:24:29 UTC, Jacob Carlborg 
 wrote:
 Why don't we instead make use of a proper framework both on 
 the server side and client side. Personally I would go with 
 Ruby on Rails but I know that most of you here would hate that 
 so a better suggestion would probably be vibe.d. For the 
 client side I'm thinking Bootstrap and jQuery.
I would suggest you avoid frameworks since they go out of fashion fairly quickly and makes maintenance dependent on individuals (with framework knowledge). jQuery adds little value since browsers are fairly standards-compliant these days, IMO.
Everybody calls everything a framework these days. Bootstrap or PureCSS are just a bunch of CSS helper classes to quickly model a responsive and modern site. I would definitely use them. Besides, any web developer worth his salt knows one or the other. And with all the tutorials they got, it is not hard to learn them either. Agree about jQuery. Its practically dead.
 It would also look very bad if you cannot run dlang.org on D 
 tech. I suggest using dlang.org for driving phobos 
 implementation/binding of standard w3 web tech. No point in 
 having D marketing other languages or their frameworks.
Yep, run the site on vibe.d, I think this is what everybody wants. Besides, vibe.d is already compiling all the html files.
 Let HTML5 layout design be a group effort and then let 1-2 
 individuals (who know what they are doing)  do the CSS styling 
 from scratch after you have the HTML5 ready.
While markup and styling are supposed to be orthogonal, in practice they rarely are. With some styling it works (like paragraphs, headers), other parts just require a coupling between html and css. You cannot just take some old HTML, apply some css styles and have a modern site. A lot of times you have to change the order, the nesting of elements or apply wrapper elements to achieve a certain look.
Jan 19 2015
next sibling parent "Ola Fosheim =?UTF-8?B?R3LDuHN0YWQi?= writes:
On Monday, 19 January 2015 at 13:43:28 UTC, Sebastiaan Koppe 
wrote:
 Bootstrap or PureCSS are just a bunch of CSS helper classes to 
 quickly model a responsive and modern site.
It is basically pointless if you know CSS and browser limitations. You need to view the site in the browsers you target anyway and add some tweaks. A programming documentation site's primary use scenario is desktop/laptop. Making it work really well there should be
 While markup and styling are supposed to be orthogonal, in 
 practice they rarely are. With some styling it works (like 
 paragraphs, headers), other parts just require a coupling 
 between html and css.

 You cannot just take some old HTML, apply some css styles and 
 have a modern site. A lot of times you have to change the 
 order, the nesting of elements or apply wrapper elements to 
 achieve a certain look.
Which is why I wrote "HTML5 layout". XSLT can transform and change order.
Jan 19 2015
prev sibling next sibling parent reply Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/19/15 5:43 AM, Sebastiaan Koppe wrote:
 Yep, run the site on vibe.d, I think this is what everybody wants.
 Besides, vibe.d is already compiling all the html files.
I think that would be great. Who wants to champion this project? Our webmaster is okay with it and can arrange a domain such as new.dlang.org to work the kinks out. Takers? Andrei
Jan 19 2015
parent reply Jacob Carlborg <doob me.com> writes:
On 2015-01-19 16:57, Andrei Alexandrescu wrote:

 I think that would be great. Who wants to champion this project? Our
 webmaster is okay with it and can arrange a domain such as new.dlang.org
 to work the kinks out.

 Takers?
I would like to but I have too many other projects going on right now. -- /Jacob Carlborg
Jan 19 2015
parent "anonymous" <anonymous example.com> writes:
On Monday, 19 January 2015 at 20:42:43 UTC, Jacob Carlborg wrote:
 On 2015-01-19 16:57, Andrei Alexandrescu wrote:

 I think that would be great. Who wants to champion this 
 project? Our
 webmaster is okay with it and can arrange a domain such as 
 new.dlang.org
 to work the kinks out.

 Takers?
I would like to but I have too many other projects going on right now.
Then STFU stop telling everybody what and how to do.
Jan 19 2015
prev sibling parent Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/19/15 5:43 AM, Sebastiaan Koppe wrote:
 Bootstrap or PureCSS are just a bunch of CSS helper classes to quickly
 model a responsive and modern site. I would definitely use them.
 Besides, any web developer worth his salt knows one or the other. And
 with all the tutorials they got, it is not hard to learn them either.
Thanks, I looked over getbootstrap.com and purecss.io and they look sensible, something I could work with.
 Agree about jQuery. Its practically dead.
I'll note that Bootstrap requires jQuery. Looks like the web design world is as opinionated as the programmers'. :o) Andrei
Jan 19 2015
prev sibling parent reply Jacob Carlborg <doob me.com> writes:
On 2015-01-19 13:00, "Ola Fosheim =?UTF-8?B?R3LDuHN0YWQi?= 
<ola.fosheim.grostad+dlang gmail.com>" wrote:

 I would suggest you avoid frameworks since they go out of fashion fairly
 quickly and makes maintenance dependent on individuals (with framework
 knowledge). jQuery adds little value since browsers are fairly
 standards-compliant these days, IMO.
These are well established frameworks used by very many developers. I would bet that each of them have more users than D has. If you don't like frameworks you free to call the tools something else.
 http://caniuse.com/#feature_sort=score

 It would also look very bad if you cannot run dlang.org on D tech. I
 suggest using dlang.org for driving phobos implementation/binding of
 standard w3 web tech. No point in having D marketing other languages or
 their frameworks.
I guess you stopped reading after "Rails" since I said it would be better to use vibe.d.
 1. semantic markup i  XML (DDoc 2 XML would be a good startingpoint)

 2. XSLT for transforming XML to HTML5

 3. K.I.S.S. design wise: WAI+HTML5+CSS3 (marginal use of JS)
I suggested well established tools for web development. I can't say that is true for Ddoc. I also never heard of anyone using XSLT for web development.
 It would be better to use dlang.org development to drive the design of
 vibe.d and phobos and front D as a capable server tech.
Again, you just stopped reading the post. -- /Jacob Carlborg
Jan 19 2015
parent reply "Ola Fosheim =?UTF-8?B?R3LDuHN0YWQi?= writes:
On Monday, 19 January 2015 at 20:41:08 UTC, Jacob Carlborg wrote:
 These are well established frameworks used by very many 
 developers. I would bet that each of them have more users than 
 D has. If you don't like frameworks you free to call the tools 
 something else.
There is no solid gravity in web development, so if dlang gets an overhaul every 4-8 years it probably is better to avoid them. As IE9/IE10 are phased out a new situation is being established on the browser side, e.g. you are getting things like Polymer (forward-looking "frameworks" driven by browser vendors).
 I suggested well established tools for web development. I can't 
 say that is true for Ddoc. I also never heard of anyone using 
 XSLT for web development.
XSLT is fine for generating static sites, but you can use anything that is decent at transforming a semantic data structure. By using XML+XSLT you market D as having a standard supporting infrastructure. That's the key issue. If you want people to move to D, you need to show that there is standard generic infrastructure that makes the transition easier. If server development in D means using hodge-podge ad hoc tech... then that is a bad sign. But if dlang.org goes for hodge-podge, then it might as well silently use Ruby/Python/Go/ whatever... since D are then not using dlang.org to position itself as a standard supporting server platform. *shrugs*
Jan 19 2015
parent "Ola Fosheim =?UTF-8?B?R3LDuHN0YWQi?= writes:
An alternative is to create a good D wrapper for Zorba. I haven't 
tried Zorba myself, but it provides XQuery3/JSONiq support and is 
written to be embeddable (C++/C).

XQuery FLWOR expressions might work out ok combined with caching. 
FLWOR is supposedly easier for many programmers than XSLT and 
provides roughly the same expressive power. Both XQuery and XSLT 
are declarative languages and XPath based.

http://www.zorba.io/
http://www.jsoniq.org/

D badly needs to show support for XML infrastructure to be taken 
seriously.
Jan 20 2015
prev sibling parent reply "Dicebot" <public dicebot.lv> writes:
Framework doesn't really matter (though dog-fooding is desirable) 
- it is a matter of maintainability. You may like RoR but how 
many people in D community will be able to maintain such solution 
if you disappear? I doubt many.

It is sad that Andrei wastes his time on such tasks but it is 
hard to find volunteers for such unpleasant things. I'd 
personally avoid it at all costs :)

This makes me wonder how much rewriting + maintaining something 
of dlang.org scale would cost if actually paid for. I'll ask some 
fellow web devs - it will be definitely much cheaper in my home 
country than in USA :)
Jan 19 2015
next sibling parent Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 1/19/15 6:54 AM, Dicebot wrote:
 Framework doesn't really matter (though dog-fooding is desirable) - it
 is a matter of maintainability. You may like RoR but how many people in
 D community will be able to maintain such solution if you disappear? I
 doubt many.
That's exactly right. That said, suggestions are nice as long as I take them as stuff I ultimately need to work with, not their proponents.
 It is sad that Andrei wastes his time on such tasks but it is hard to
 find volunteers for such unpleasant things. I'd personally avoid it at
 all costs :)
I did the same, with the result that we all know. I figured I can't afford the luxury of avoiding certain work.
 This makes me wonder how much rewriting + maintaining something of
 dlang.org scale would cost if actually paid for. I'll ask some fellow
 web devs - it will be definitely much cheaper in my home country than in
 USA :)
That would be interesting, thanks! Andrei
Jan 19 2015
prev sibling next sibling parent "Ola Fosheim =?UTF-8?B?R3LDuHN0YWQi?= writes:
On Monday, 19 January 2015 at 14:54:36 UTC, Dicebot wrote:
 It is sad that Andrei wastes his time on such tasks but it is 
 hard to find volunteers for such unpleasant things. I'd 
 personally avoid it at all costs :)
Uhm, if you insist on home grown macros, and provide no consistent semantic markup in a standard format and set up for a process where bikeshedding the visual design is the norm... then there is little room for volunteers. If you want volunteers who know the topic... plan for a pleasant process. Self-inflicted damage... :-P
Jan 19 2015
prev sibling parent Jacob Carlborg <doob me.com> writes:
On 2015-01-19 15:54, Dicebot wrote:
 Framework doesn't really matter (though dog-fooding is desirable) - it
 is a matter of maintainability. You may like RoR but how many people in
 D community will be able to maintain such solution if you disappear? I
 doubt many.
I said like 100 times now that vibe.d would be a better choice but I guess everyone stopped reading after "Rails". -- /Jacob Carlborg
Jan 19 2015