Hi Caty,
Just before giving some arguments below, let me say that I'm very happy that
you're working on
and I globally like very much what you're doing, no
problem there :)
On Sep 27, 2013, at 7:22 PM, Ecaterina Moraru (Valica) <valicac(a)gmail.com> wrote:
Hi,
Thank you for your votes. I've integrate the Homepage changes and we are
using the Junco skin on
xwiki.org
__Discussion 1__
So one of the problems is 'Clearing your Cache'. In order to see the
changes correctly you need to refresh the page and make sure the cache is
updated with the new style rules.
It would be great if someone would know how we could force the clearing of
the cache. Maybe
http://jira.xwiki.org/browse/XWIKI-6073 ?
__Discussion 2__
On a related topic, Vincent thinks we should not use a 'centered' design,
instead have (just like Colibri has) a 'full width 'design.
TBH I don't know yet what is better yet but I'm asking myself and the comments
below are my comments only from first sight. I'm sure I can adapt or have a different
opinion given time :)
Why I disagree (having full width and prefer the
centered version):
-
xwiki.org is a documentation site so most of its content needs to be
read;
Centered is nice for websites, ie site for which you control the content and can lay it
out nicely. This is not really our case and it's hard to have something that looks
nice for all pages.
- From a readability point of view, if the line is too
long it is harder
for the eye to read it, please read
http://baymard.com/blog/line-length-readability ;
I haven't read this link yet but what I can tell you is that right now I have a hard
time reading content because:
* I see only a really small portion of it at once
* The font is quite larger than before
I don't know yet if it's better or worse than before but I certainly need some
adaptation time.
I think the main issue is that our UI is too crowded for having a centered + large font
UI.
For example I like this site quite a lot (Note that they use the full width too):
It's large too but the interface is very clean and I find it nice. In our case we have
various sizes everywhere (look at menu sizes vs content size) and lots of eye distraction
with all the menus.
- Being a responsive skin, for smaller devices the
content goes full width,
so the centered width is just for large resolutions;
Let me know what you think,
Caty
On Thu, Sep 26, 2013 at 11:08 AM, Ecaterina Moraru (Valica) <
valicac(a)gmail.com> wrote:
>
>
>
> On Wed, Sep 25, 2013 at 11:13 AM, Ludovic Dubost <ludovic(a)xwiki.com>wrote;wrote:
>
>> Great job indeed. I installed on a test wiki on a 4.5 farm and it looks
>> quite nice.
>>
>> I have a few questions.
>>
>> 1/ If I understand properly we have a legacy.css where lives all CSS of
>> HTML that would not have been made clean plain bootstrap compatible HTML.
>> Is that so ?
>>
>
> I've took colibri.css:
> - stripped it from any rules that Bootstrap already covers (reset,
> standard html elements, forms, etc.);
> - mapped what I could to Bootstrap components (tables, boxes, panels,
> etc.);
> - stripped layout elements and tried to get rid of ColorThemes values (in
> the end realizing that I can't, without loosing functionality);
> - cleaned the remaining selectors and properties (removed IE6,7 rules;
> removed rules that maybe are covered by Bootstrap; etc.)
>
> What remained are classes that will be needed by any new skin; classes
> that kind of represent the CSS API (although there are still parts that
> could be moved to their specific component file).
> I though that if someone makes a change in Colibri and that selector is
> also in legacy.css it should be ported here (that's why I left the
> selectors as original).
>
> Everything that is in legacy.css could be rewritten in junco.css, but it
> will create a bigger gap between the two skins.
>
>
>> 2/ In terms of work, how much and what is needed to get to a point where
>> this could become our default skin, knowing that we should not have
>> feature
>> regression, for instance users should still be able to build their color
>> themes in Wysiwyg like they can with colibri.
>>
>> I'm sure we can find a solution to do on the fly compiling when saving
>> color themes so to solve the issue of needing LESS.
>>
>
> The integration problems are presented above (__Platform integration
> problems__). One is the need to be able to build LESS files and update them
> 'on the fly'.
> Regarding ColorThemes values, there are 2 files:
> - global/xwiki-colorthemes.less - which is kind of a colorThemeInit.vm
>
>
https://github.com/evalica/bootswatch/blob/junco-themes/global/xwiki-colort…
> - <theme>/xwiki-colortheme.less - which overrides the initialization and
> provide colors for a specific theme
> Blueberry example:
>
https://github.com/evalica/bootswatch/blob/junco-themes/blueberry/xwiki-col…
>
> So when changing ColorThemes values, we need to make sure we update this
> files.
>
>
>>
>> 3/ How much is needed to get rid of legacy and have a fully native skin
>> with the new system ?
>>
>
> As stated above, legacy.css is not a limitation. We could rename it from
> legacy.css to base.css or api.css, we could maybe write some more
> performant selectors or maybe write them using LESS nested selectors.
> The file contains the common denominator for Colibri and Junco and some
> parts of the layout could be removed from it, in order to make it a true
> API file. All the selectors that will be removed from here will end up in
> the xwiki-common.less
>
https://github.com/evalica/bootswatch/blob/junco-themes/global/xwiki-common…
>
> If we rewrite it whole, the only problem is that it will be harder to
> update both Colibri and Junco.
>
>
>>
>> 4/ What is the migration path for a wiki where a custom skin has been
>> built
>> on colibri
>>
>
> Because of the mapping between XWiki and Bootstrap the Colibri classes
> will be covered in Junco.
> The problem are base HTML elements, that may have other default values.
> I'm sure tables will produce some problems, because in Colibri they didn't
> had any default padding, while after mapping they do. If someone use table
> for layout, there will be some extra spacing.
> There is not migration path for skins like that. Although we should have
> minor problems, they will need to be tested and fixed manually.
>
>
>>
>> 5/ What are the potential consequences on future compatibility with
>> colibri
>> based skin, particularly if we start modifying our HTML produced by
>> different modules ?
>>
>
> So the current HTML structure is covered by legacy.css and
> xwiki-mapping.less.
>
> If someone will want to change the HTML structure, you need to:
> - make sure that the classes you delete are not 'API', this means they are
> not called in legacy or xwiki-mapping. If a class is present in these files
> means it also has a significance in Colibri;
> - if you find a certain class, don't delete it. Instead just append your
> new class or Bootstrap class. Keeping the legacy class, it will have a
> fallback if Colibri skin will be used.
>
> Of course there will be differences between skins if we start using
> Bootstrap components and that is normal (because Colibri will not know of
> their existence).
>
> Thanks,
> Caty
>
>
>>
>> Ludovic
>>
>>
>>
>> 2013/9/20 Ecaterina Moraru (Valica) <valicac(a)gmail.com>
>>
>>> Hi devs,
>>>
>>> For the past weeks I've been working on a skin based on Bootstrap[1].
>> You
>>> can read more about it and test the new Junco Skin at
>>>
http://extensions.xwiki.org/xwiki/bin/view/Extension/Junco+Skin
>>>
>>> This proposal is about using the Junco Skin (Blueberry
>>> Theme<
>>>
>>
http://extensions.xwiki.org/xwiki/bin/view/Extension/Junco+Skin#HTheme:Blue…
>>>> )
>>> for
xwiki.org.
>>> I prepared some responsive screenshots for the
xwiki.org Homepage and
>> an
>>> Extension page.
>>>
>>>
>>
http://incubator.myxwiki.org/xwiki/bin/view/Improvements/XWikiOrgJuncoHomep…
>>>
>>> This is my +1
>>>
>>> Please report any issues on
>>
https://github.com/evalica/bootswatch/issues
>>>
>>> __Advantages__
>>> - a change is always welcomed, shows the users there is activity on the
>>> website;
>>> - the skin is responsive;
>>> - by using Bootstrap we have the whole framework's power to use (grids,
>>> components, etc.);
>>> - we have the chance to test a bit the skin in production and see the
>>> possible bugs, in order to later integrate;
>>> - IMO the skin looks nice :)
>>>
>>> __Disadvantages__
>>> - the only disadvantage is that there will be bugs and they will take
>> some
>>> time to be detected and fixed;
>>>
>>> __Platform integration problems__
>>> 1. currently the new skin uses the HTML5 doctype. This is needed if we
>> were
>>> to use some Bootstrap JS components (carousel, menus, etc.) - which we
>>> currently don't on
xwiki.org, so I could use the old doctype (but lose
>>> some
>>> of the testing purpose). Because of the HTML5 doctype, the HTML
>> validation
>>> fails. See
http://jira.xwiki.org/browse/XWIKI-7552
>>>
>>> 2. Junco Skin currently doesn't have support for changing the
>> ColorThemes
>>> on the fly. I would need the help of a developer to fix this problem. Is
>>> not so much a problem for
xwiki.org (I could fix them by duplicating
>> some
>>> code in the
xwiki.org skin), but it is a problem for the integration.
>>>
>>> 3. Being build on Bootstrap, it needs LESS to compile the files into
>> CSS.
>>> Again I would need a developer to see how we could integrate the
>> building
>>> of the themes in platform. Right now this is done locally, partially
>> manual
>>> by using Grunt.
>>>
>>> Tell me what you think and take some time to test the skin,
>>> Caty
>>>
>>> [1]
http://getbootstrap.com/
>>> _________________________