31 May
2011
31 May
'11
3:38 a.m.
On 05/26/2011 11:48 PM, volker wrote:
I want to give feedback to the Xwiki concept.
Hi Volker,
Thanks a lot for the feedback.
As a radiologist I am searching for a software that do
three simple
tasks:
1. Storing anonymous radiological images for teaching purpose.
2. Connect facts, notes,articles (text, PDF, PTT and JPG to this
images.
3. Using a system to retrieve the information.
Why? I often know that I have seen the phenomena/disease before. But I
do not remember where I found facts to this phenomena/disease. So why
searching for the same disease a second (10th) time?
Se: http://dx.doi.org/10.1594/ecr2011/C-2126
Nice, thanks for the reference.
Xwiki has a REST interface. I designed a universal
image import
macro(with autohotkey.net) that grab an image, open a tiny input
interface for tags and drop the image and tags the Xwikiserver.
Xwiki handle the image workflow quit well. Storing articles (PDF) to
this image is working. Tagging and hierarchical connection is working.
Lucene is doing the search job.
So what is not working?
STARTING WITH XWIKI
I am not a experienced programmer nor a database freak. Working before
with twiki.org and mediawiki.net on PHP. Never working with Java server
before.
So I started from a USB device with the Xwiki stand alone server on
Windows XP. After some weeks I realized that IT IS ONLY FOR TESTING!
(Yes it is informed on the page).
Indeed, the default configuration is suited only for testing,
sacrificing memory for speed.
MIGRATING TO MYSQL
Using “MoWeS Portable” freeware for Apache, MySQL, phpMyAdmin.
Export from “testing” and import to MySQL is not working for me because
of large database and umlauts in page name and attachments. The
ImportLargeWiki script is working after manual converting all umlauts
and question marks from page name. Se next section.
UMLAUTS
I am working in Norway with not updated Windows XP and Swedish code
pages. Developing at home on a German Ubuntu: In this setting it is not
possible to generate a page with page name, page title and attachment
name including German or Norwegian umlauts.
1. The page name must not include umlauts or special characters.
2. Page title is safe with umlauts or special characters.
Any character should be supported in the page name. If it doesn't work,
then there's a configuration error somewhere. If you want, we can have
separate discussion to try and fix this.
3. Attachment name must not include umlauts or
special characters .
This is a limitation that should be fixed soon.
It is not optimal, but I can (now after intensive
testing) live with it.
I use a time stamp for file name and for attachment name.
IMAGE WORKFLOW
My automated import workflow is quit nice, so I used it many times. The
upload and search for image dialogue is using the page title and is
working god.
I still miss the possibility to make a image annotation. (like in
Wikipedia) and the link to full size image.
Linking should be possible. See for example the second image on
http://www.xwiki.org/xwiki/bin/ReleaseNotes/ReleaseNotesXWikiEnterprise30
Image annotations will be implemented sometime in the future.
Why not have a thumbnail of the image in search result
page and
AllAttachment page?
This should be doable, please open a couple of issues on jira.xwiki.org
But the images in the database are blowing up the size
of the database.
Performance is going down, handling import and export is getting time
consuming.
Starting with XWiki3.0 you can use the Filesystem attachment store, see
http://www.xwiki.org/xwiki/bin/ReleaseNotes/ReleaseNotesXWikiEnterprise30#H…
for more details
ATTACHEMENT WORFLOW
Missing a search for “foo bar “ attachment in other pages/spaces.
I'm not sure I understand this.
LUCENE SEARCH
My search results are giving 10jpgs first and then the parent page. (It
is a known bug).
Yes, it is a known bug which I was supposed to fix in 3.1, but didn't
have time. It's planned for 3.2 now.
Id like to see fragments of the document on the result
page.
Highlighting the search text is nice. Is it a Xwiki issue or is it easy
to realise a show text on the result page?
It's a planned improvement, but it's not in the near-future roadmap yet.
Show me the images as thumbnails and PDF/ppts as
icons.
Image thumbnails is easy to implement, PPT previews should be doable as
well, but PDF might take some more time.
How to change the Analyser? Is there a Norwegian text
analyser for
lucene? (I know it is not a genuine Xwiki issue, but helps to make Xwiki
more efficient)
Changing the analyzer is rather easy, but there's no Norwegian analyzer
for the version of Lucene that we're using. There's a new one in Lucene
3.1, and we're planning on upgrading to that version very soon (probably
in XWiki 3.2 as well).
EXPORT PDF /PRINT
Tables are not printed with border. The image positioning (right-left)
seems to get broken in PDFs
You can create a file called pdf.css in xwiki/templates/, with this
content in it:
table {
border-collapse: collapse;
}
table th, table td {
border: 1px solid #000;
}
RELAUNCH
Now with Xwiki 3-1.2. On the Norwegian XP with Swedish code page.
1. Plain Xwiki installation with FILE system
2. MySQL via MOWES. Just start MyPhpAdmin to add xwiki database
3. No Login on plain XWIKI!!!
This is normal, since the database is empty, and all users are defined
in the wiki. The guest user has full admin rights, though, so you don't
need to log-in to perform the next steps.
4. Plain 3-1.2 XAR installation.
5. Install script LargeWikiImport from Xwiki.org
6. Export of XAR from the old system with ExportLargeWiki by
spaces.
7. Import with LargeWikiImport. USING “do not overwrite” option
8. Give time to the system and lucene search.
9. One problem. All the old pages are “hidden” (browse tables with
MyPhpAdmin)
1. Probably because used “backup”- option on import. (Is
this right? Correct me if wrong. AND PUT SOME INFO IN
THE SCRIPT)
No, this looks like a bug in the configuration. What version of the
mysql connector jar are you using? What version of MySQL does the MOWES
application include? XWiki requires a 5.x version both for the connector
and the database.
2. In MyPhpAdmin update from hidden=1 to
hidden=0 for all
pages
And now: A fast system with newest software version, working from a USB
device on Windows and Linux.
But there is a maximum file per directory bounder on Windows systems.
FAT16- 512
FAT32- 65,534 files or folders per folder
NTFS- 4,294,967,295
Is this correct?
MISSING
I want to include pages with sections to make a book. The hiracical
section including is not working well.
What do you mean?
Lucene: The search syntax shoot be available inside
the wiki.
Indeed, we could at least add a link under the search bar to the Lucene
syntax guide.
Why is the maximum upload parameter so low? Put it up
to 1GB. And make a
link to the edit page. Not “se in forum and go back to the Wiki”. It is
confusing.
When attachments are stored in the database, trying to upload large
files can break the wiki very fast.
How to migrate users?
Users should be in the XAR as well, in the XWiki space.
Why not putting all Xwiki system pages in a space. And
only the WebHome
in Main?
Most system pages are already in the XWiki space, and only some public
pages are in the Main space. Which documents do you think should be
moved as well?
On upgrading/migration process you upgrade only
(first) space Xwiki. And
in a second step you copy all your pages from Main to (empty) Main.
Why not using LargeWikiImport and LargeWikiExport as default?
Improving the import and export is also in the near roadmap (3.2 or 3.3
as an estimate target).
I develop and use Xwiki for RadWiki (Radiology Wiki)
on a daily base
now. It is fun to use it!
Thank you for your attention. Volker
Thanks again for the feedback. The first step towards fixing issues that
annoy you is to bring them up on the list or the issue tracker.
--
Sergiu Dumitriu
http://purl.org/net/sergiu/