Hi,

this is just a wild guess, put let's put ourselves in diffmk shoes,
walking down the XML tree:

1- I find a title, identical content, OK
2- I find a para, identical content, OK
3- I find a para in "old", it is also in "new", I output it.
4- When I compare the content of this para I realize it is different, so
I -Huh- do something
5- Next para is only in "new" so I<para diffmk:change="added">

Providing diffmk does not "look forward" -which would have allowed it to
see para2 was actually moved- I think the issue is in step 4 where we
should have "@diffmk:changed = 'changed'".

Actually it's not the first issue I found in diffmk, Norman always
stated this was alpha, and I'm afraid he has other xml cats to whip at
the moment...

XML diff is quite a challenge, some guys spent PhD thesis on the topic...

Camille.

Jeff Hooker wrote:
> Hi,
>
> After more testing, I've found an issue which I hope someone finds familar.
>
> For in-line changes, DiffMk works fine, wrapping changes in tagged markers 
> exactly as advertised. For changes to structural elements, however, it is 
> reliably incorrect. When adding a PARA element, for example, it will wrap the 
> content of the PARA element in a new element, and then also mark the 
> following sibling PARA element as "added". This holds true for sections, 
> chapters, books, etc. Adding one element results in the following sibling 
> being marked as "new.
>
> old.xml
> -------
> <article>
>     <title>This is the title</title>
>     <para>para1</para>
>     <para>para2</para>
> </article>
>
> new.xml
> -------
> <article>
>     <title>This is the title</title>
>     <para>para1</para>
>     <para>para3</para>
>     <para>para2</para>
> </article>
>
> output.xml
> ----------
> <article diffmk:version="3.0" xmlns:diffmk="http://diffmk.sf.net/ns/diff";>
>     <title>This is the title</title>
>     <para>para1</para>
>     <para><diffmk:wrapper diffmk:change="added">para3</diffmk:wrapper></para>
>     <para diffmk:change="added">para2</para>
> </article>
>
> The problem affects only the following sibling; if the element is that last 
> child element, then the next elements on the tree are unaffected. If there 
> are more than one siblings following the affected element, they are untouched.
>
> Does this look familar to anyone?
>
> Thanks,
> Jeff.
>
> -----Original Message-----
> From: Camille Bégnis [mailto:[EMAIL PROTECTED]
> Sent: Friday, May 30, 2008 2:15 AM
> To: Jeff Hooker
> Cc: apps docbook
> Subject: Re: [docbook-apps] DiffMk configuration
>
>
> Hi,
>
> please find attached those two files as I use them, I can't recall if I
> modified them for my own use, better review it.
>
> Hmmmm after digging further I think Norm had sent me a modified jar, I
> might have modified the XSL too, so I put the whole modified package I
> use here: http://www.neodoc.biz/files/diffmk-3.0.a1.patch.zip
>
> Note I could never make the --diff option work.
>
> Good luck,
>
> Camille.
>
> Jeff Hooker wrote:
>   
>> Hi all,
>>
>> I *think* this is appropriate for this list. If not, my apologies.
>>
>> I'm trying to get DiffMk (http://nwalsh.com/java/diffmk/index.html) up and 
>> running using Bob Stayton's instructions 
>> (http://www.sagehill.net/docbookxsl/Changebars.html) and not having much 
>> success for far.
>>
>> At the moment I'm just getting "Failed to load Main-Class manifiest 
>> attribute from diffmk.jar" but I suspect that this is a config issue rather 
>> than a java issue.
>>
>> Bob's instructions include reference to two files that do not seem to exist 
>> in the current diffmk applicaton: diffmk.xml and DiffMk.properties, both of 
>> which are supposed to reside in a nonexistant directory called "config". 
>> I've tracked  down a sample DiffMk.properties file, but am working blind 
>> with the diffmk.xml file.
>>
>> Has anyone got this thing working, and if so, could you send me some 
>> guidance?
>>
>> Thanks,
>> Jeff
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [EMAIL PROTECTED]
>> For additional commands, e-mail: [EMAIL PROTECTED]
>>   
>>     
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>   
begin:vcard
fn;quoted-printable:Camille B=C3=A9gnis
n;quoted-printable:B=C3=A9gnis;Camille
org:NeoDoc
adr:;;139 rue Philippe de Girard;Pertuis;;84120;France
email;internet:[EMAIL PROTECTED]
tel;work:+33.9.54.96.99.55
tel;fax:+33.9.59.96.99.55
tel;cell:+33.6.33.15.10.23
url:http://www.neodoc.biz
version:2.1
end:vcard


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to