Hi,
Since we introduces ColorThemes in XE 2.0 we've used them constantly, being
a great improvement for the cases when we wanted to rapidly change the
color variables used.
The problem with the ColorThemes is that they contain a limited number of
variables. There are places in Colibri where we've recycled and used
variables that were not intended to be used there, for example:
- Administration's vertical-menu is using
$theme.panelCollapsedBackgroundColor (which was intended for the collapsed
state of the panels);
- Suggest search results is using for it's .resultContainer also
$theme.panelCollapsedBackgroundColor;
- etc.
Also there are some variables in the ColorThemes like
$theme.textPrimaryColor and $theme.textSecondaryColor or even
$theme.backgroundSecondaryColor that are used quite randomly in the code
(all being shades of gray in the default theme). Not knowing exactly what
is the purpose and usage of a color is very easy when you create a
ColorTheme to chose colors that in practice will not provide enough
contrast or that will not look very nice together.
Since their release, we've added just 6 new variables (2 for the "Add" menu
and 4 for the notifications colors: warning, error, success and info). One
of the reason behind not adding new colors was not to confuse the user with
a wide range of color variations. The colortheme wizard was great at making
it easy for the user to change the 36 variables, but having 50+ variables
to set could be a difficult/complex task even when using a Wizard.
Although when we created ColorThemes we had in mind that they will be skin
independent I don't know if in practice this will be still valid. When
we're going to create a new skin for XE it will have a different layout, a
different semantic of the variables names and a different color pool need.
These needs are not mappable on what we currently have.
So I think the ColorThemes should be skin dependent and also version
dependent in order to assure that is gonna look the way it was originally
intended to look.
For example I want to add some CSS3 gradients for the 3.4 Colibri skin, but
I don't have enough variables to specify the gradient shades. To solve that
I've used non semantically variables combinations like:
background-image: -moz-linear-gradient(top, $theme.linkColor 0%,
$theme.buttonPrimaryBackgroundColor 100%);
background-image: -moz-linear-gradient(top,
$theme.pageContentBackgroundColor 0%, $theme.buttonSecondaryBackgroundColor
100%);
Even though the combination will work for the ColorTheme defined specially
for XE 3.4, this means the other existing ColorThemes will be used in a way
they were not designed for (like ColorThemes.Bordo theme, etc.) thus not
being as good looking as before when applied on the new skin.
The same use case will be for any other skin that we will create for the XE
4.x. IMO you can't assume that any defined colortheme will work
"graciously" independent of the skin applied and version used. (they will
work but the color mix will be random).
So another solution would be add more colors to the ColorTheme. For
variation of Colibri this will work, but for a new skin (that maybe will
not have 2 sets of menus, or will not have 2 separate colors for the
background and the content) all the semantic behind the variables naming
will fail. We will end up with variables inside the colortheme that may
never be used inside the new skin code and I don't even imagine how the
ColorTheme Wizard would look like (a nice idea would be to adjust itself
dynamically dependent of the enabled skin).
Also I want to know what is your opinion regarding adding new variables.
What is the backward compatibility strategy for that? Right now, we have a
fallback ColorTheme that provides fallback colors in case the ColorTheme
doesn't have those colors set. But this does not assure that installing an
"old" red colortheme will look nice with the fallback blue colors. We are
not telling the user anywhere that there are missing colors in the
installed colortheme and also in the wizard is hard to know what colors are
defined and what are the fallback ones (if you know what colors are missing
than you can manual define them and you don't need to cover all wizard
dialogs in order to find the inconsistencies).
So IMO the only solution is to have ColorThemes designed dependent of a
skin and a version, with the possibility to let the user manually enhance
and reuse old colortheme to work on the new versions.
Maybe you have other solutions.
Thanks,
Caty
The XWiki development team is proud to announce the availability of
XWiki Commons, XWiki Rendering, XWiki Platform, XWiki Enterprise and
XWiki Enterprise Manager 3.3.
Following the goals established for the 3.x cycle, XWiki Enterprise 3.3
delivers the first usable but experimental versions of App Within
Minutes and Extension Manager features. The highlights of this release are:
* Experimental "App Within Minutes" feature
* Improved Extension Manager
* Automatic external link checker
* Better support for exporting CJK documents as PDF
* LDAP user membership improvements
* Attachment handling improvements
* Debian packages for installing XWiki
See the full release notes at
http://www.xwiki.org/xwiki/bin/ReleaseNotes/ReleaseNotesXWikiEnterprise33 for
more details.
--
Sergiu Dumitriu
http://purl.org/net/sergiu/
Hi devs,
Here's something I'd love to see:
The idea is the following:
* Allow end users to install the extension and run the tests on their current XWiki Enterprise instance
* At the end of the test make the extension publish the result by POSTing the result to a xwiki.org URL. The results should include:
** XE version
** DB version
** DB driver version
** Browser version
** Console logs
** Screenshots taken
* Display test results on several places of xwiki.org:
** On a general compatibility page
** On the Installation page for specific Servlet Containers and Databases
Note that we need to improve a bit the test suite so that:
* All tests create data in a Test space
* If a test requires a default page of XE we need to verify it contains what we need before the test starts and if not then mark the test as skipped instead of failing it
* Tests work independently of what data already exists in the wiki
I've entered it as a jira here: http://jira.xwiki.org/jira/browse/XE-1064
WDYT?
Thanks
-Vincent
Hi Devs,
I am looking at now using the new Locale added in DocumentReference into
the implementation of XWikiDocument.
I have already deprecated language related stuff in XWikiDocument, and I
have introduce a XWikiDocument#getLocale and an XWikiDocument#isTranslation
helper since the deprecation of defaultLanguage will increase the need of
it. I have also added XWikiDocument#getTranslatedDocument() with Locale in
place of language. All the changes are almost backward compatible, which is
nice (there is some subtleties with default, "" and null that is now more
equivalent, but should not have consequences).
The is however one change that is not backward compatible, which is the
change of the document reference. Therefore,
XWikiDocument#getDocumentReference does not return the same reference than
it does before, because this reference now contains the Locale. This cause
breakage in several places. I see some option to fix this:
A. Fix all places broken. This may be too long for me, and not trivial.
B. Introduce a new XWikiDocument#getDocumentReferenceWithLocale() and
have XWikiDocument#getDocumentRefence() returns without Locale. Very easy,
but not nice.
C. Introduce a new XWikiDocument#getDocumentReferenceWithoutLocale() and
change all existing calls to XWikiDocument#getDocumentRefence() in platform
to use this one. Nicer, but this is not fully backward compatible.
Since I am on it right now, I would appreciate your opinion quickly.
WDYT ?
I am undecided between B and C
--
Denis Gervalle
SOFTEC sa - CEO
eGuilde sarl - CTO
Hi devs,
Right now, when you create a wiki template from a xar, the import that is
done in the background is a backup import, meaning that the last author of
the pages that get imported in the new wiki keep the author specified by
the xar. This often creates problems like:
- Missing Programming Rights
- Unregistered macros
- Malfunctioning scripts
These problems can appear because the user specified in the xar (even if it
is XWiki.Admin) is almost always a local user and subwiki local users do
not have PR.
If it's not a PR issue, then the user specified in the xar can be
non-existent and this makes admin checks fail, thus failing wiki macro
registration for the entire subwiki.
We are currently experiencing this problem in Workspaces, since, at the
install step, we create a workspace template by using a
workspace-template.xar (default XE but can also be user provided). Since we
make sure to delete any local users (including XWiki.Admin), the Wiki
macros will not be registered in the template and, obviously, neither in
any created workspace.
I`m hoping to include this in 3.3 final so that Workspaces can avoid the
macro registration problems (and possibly others).
So I`m asking for your vote to change the current default to non-backup.
This means that all the pages in the new subwiki template will have the
current admin user that created the template as last author.
Here's my +1.
Thanks,
Eduard
Hello,
Doing a translation I found this message to translate
{0,choice,0#No|1#One|1<{0}} included
{0,choice,0#documents.|1#document:|1}
So I wonder is it possible to use arithmetic expression like {0}%10 in
those choice statements because in my language, the noun's ending depends on
the last digit in a number .
Directing me to the page describing full syntax for the language above
would be enough.
Regards,
Roman
--
View this message in context: http://xwiki.475771.n2.nabble.com/Use-complex-expressions-in-i18n-choice-tp…
Sent from the XWiki- Dev mailing list archive at Nabble.com.
Hello,
I've realized recently that panel titles are exposed in h1. For example :
<h1 class="xwikipaneltitle">Quick Links</h1>
I don't think that's the best we can offer semantically. Panels are
not very import sections of the page. Maybe h2 would be more
appropriate ?
WDYT ?
Jérôme Velociter
Winesquare
http://www.winesquare.net/
Hi devs,
I have just installed the XEM distribution has an upgrade of an existing
farm.
I have discovered that the new distribution cause an new DB to be created
without any confirmation on first access to Main.WebHome to contains the
workspace template required by the new workspace feature. IMO, this does
not follow the vote we have had before about this change, since existing
user of XEM are immediately impacted by the new feature. I would have been
-1 if I would have been aware of this.
Creating the new database this way has for me some inconveniences:
- For new user, after having a possibly hard time setting up the
server, there first successful access on the wiki could end in an error,
since the creation of the new database could goes wrong.
- For existing user, a new database is created without there agreement on
there farm. If they delete it, it will probably came back again and
again... They really need to understand that the simple access to the home
page cause the creation of a new database.
IMO, these are not good first experiences with the new release. I propose
that the creation of the new database requires at least a user
confirmation, and that if the user do not confirm, it should not be tried
anymore. This will have two advantage, the user is well aware that a other
new database is under creation in case of error, and existing farm user
will have an easy way out.
WDYT ?
--
Denis Gervalle
SOFTEC sa - CEO
eGuilde sarl - CTO
Hi,
I just tried the XEM snapshot and got an error when trying to configure it.
I get this after login as Admin:
Workspace Manager translations have been successfully installed.
Workspace search suggestions have been successfully installed.
Error installing workspace template. null
Workspace template features can not be installed without the workspace base
template being installed first.
I also agree with Denis that any install should be done with a confirmation.
Ludovic
--
Ludovic Dubost
Founder and CEO
Blog: http://blog.ludovic.org/
XWiki: http://www.xwiki.com
Skype: ldubost GTalk: ldubost