On May 27, 2010, at 6:24 PM, Jonathan y Yolaine wrote:
> > Hi Steve,
> > I like how you made the item generic both for BSW and for other languages. It makes for a flexible structure that would even support sign languages back to back.!!! Also this way it would support unlimited terms or glosses for each entry.
I'm glad you like SPML. I still to make a few tweaks, but I like it
too. One thing I really like is the ability to go from markup to
database. I created an SPML to SQL script. Next I'll need to reverse
it so I can go from SQL to SPML. You can see both the SPML and the SQL
for the English/ASL User Interface:
> > Is SPML an exchange format just for the dictionaries or also for the literature?
The SPML exchange format will be used for both. The <entry> element can
have 3 literature oriented attributes: top, previous, and next. These
provide the linking between entries and will be used for literature with
multiple pages. Eventually, the markup will probably need to be
customized to differentiate dictionaries from literature.
> > There are 5 things that I would like to suggest.
> > 1. Add a Global unique ID (GUID) for each entry so that it will be easy to compare whether we are exchanging an entry that we already had. The "entry id" good for the SignPuddle database, other programs may not have the same id in their database. The GUID permits you to detect whether we are talking about the same entry or not.
A global unique ID may be useful and can be constructed right now.
There are 2 main types of puddles: ui and sgn. Each puddle has a unique
ID. So entry 7 is the user interface puddle number 1 would have the
unique ID of "ui.1.7". I need to add the puddle type and id as
attributes to the <spml> tag.
> > 2. Add a time stamp to each entry, so that we can compare to of the same signs and help decided if we should update the one we have, or not change anything if the sign is older or the same as the one we already have.
This is already there. Entries and items have 2 dates: CDT (created),
and MDT (modified). Unix time stamps are used.
<entry e_id="2" cdt="1172438830" mdt="1173731285">
<item i_id="3" lang="ase" cdt="1172438830" mdt="1173731285">
<item i_id="4" lang="en" cdt="1172438830" mdt="1173731285">
<term t_id="4"><![CDATA[User Interface]]></term>
<text><![CDATA[Contains the text used for the display.]]></text>
> > 3. Add an identifier to the item tag to specify whether it is a BSW string or other foreign string. Of course you could always use the an algorithm to determine this but when you have a really long list it is much faster to read the value that to decided which it is for each one.
Interesting idea. I'm still mulling this over.
> > 4. It seems to me that the "src" tag should belong to the entry tag and not to the item tag. It is repeated many times. You may have other reasons for this structure which may not be apparent to me at this time.
There are 2 reasons to repeat the source. First, one person can add an
ASL sign and another person could add the BSL. Second, one person could
add an ASL sign and another person could add a different ASL sign.
> > 5. If you are only going to have one name tag inside the item tag and the item tag doesn't have any other children, then the name tag is redundant. Though it might add to legibility.
Is it easier to be redundant with a simple definition without special