>>>Hmm, I missed the bit about the HTML5 persistent storage PR blitz, is there an executive summary somewhere?
>>
>>Google for the decision of Firefox not to support WebSQL and the day job of the committer who
>>gave them the IndexedDB on BerkelyDB (Hint: who owns it ?)...
>>I am sooo glad they saved me from the dangers of using a single SQL caching client layer
>>on the probably most used database tool in the world (which is used in Tunderbird for neraly all stuff)
>>and try to force me into ORM Key-Value rubbish on disk as well.
>>
>>Yuck. Hope many Oracle/SQLServer licenses will be exchanged/saved by going NoSQL in backends
>>as there is better/more natural support for such structures in browsers now.
>>
>>>UPDATE: there seem to be a couple of APIs, "Web SQL Database" and "Indexed Database API". The former seems to be moribund, don't know about the latter. The first articles I've seen about them discuss SQLite rather than MySQL.
>>>
>>>Side issue, since you mentioned web-based being "noticeably slower than native apps": I wonder what is considered "fast enough", these days. Years ago I remember hearing that a response time of 2 seconds was considered "fast enough". It's probably faster today. After all, we have a civic duty to reduce car crashes etc. due to inappropriate apping/texting etc. :-/
>>
>>Target: 100ms, with outliers falling under 200ms that is for a "reactive app".
>>Otherwise 1 sec to keep train of thought (or lowered attention span of today's kids).
>>After 7-10 secs user will work on something else.
>>Opinioated Source: Google in talks on optimizing GWT JAVA>>JS Compiler and Minifier
>>
>
>Have you done anything with NoSQL? I have read some articles about it in MSDN Magazine and elsewhere and have to admit it's a mind bender for me. Maybe I am just too locked into the SQL way of doing things.


depends on your definition - no if you mean pure NoSQL databases.
I had late last century email exchange over more than 9 months with Carl Rosenberg
- IIRC his name correctly -, as he planned to build an object data store on new ideas
and wanted to draft me into this venture as I had a nearly unencumbered SQL interpreter
which was partially optimized on dBase table data (created for a ModulaII/C data library tool set.

Turned him down in the end - he created the db4o database, one of the more prominent
players working in the free/low cost range and his company was bought up.
Might have been a mistake in retrospect, but unsure if had stayed that long -
and I was already in planning for a very lucrative job back then.

Back then I saw some benefit in object data, but I believed then and now that object data
should be only be a subcategory of normal data. I had already hat great success with vfp's
Array to Memo and later used Ini,XML and JSON structures in memo fields.
db4o is a pure object database - and got great reviews after ~2004.

So I am not so totally pureSQL, but for my data sizes normal data backends always were
sufficient (no Hadoop necessary) and I scratched the perceived need with the tools at hand.
Started entering structured data well before XML was added onto SQL server -
even if normalization afficionados shudder, it made sense inside the project.

regards

thomas