mORMot Open Source

AOG,

Maybe there is a slight misunderstanding here. What I  meant was creating the  code to interface OpenUI5 with mORMot in pascal using SMS. It could be a better fit for Delphi/FPC programmers who are recluctant to get into the depths of JS. It could make locating problems easier or adding their own work.

AOG,

You seem to write a lot of JS code. Would it not be better to use SMS? It would allow Delphi/Lazarus programmers to join.

Anyway, Great work!

Leslie

AB,

Do you consider SMS to be stable enough to be used with mormot? (The none UI part)

I have no control over the php code on the server side. I just need to access  the data in the database created by php serialization in my pascal code.

I am looking into mORMot's serialization if there is an easy way to create a custom serialization for this. Some guidelines/examples could be useful.

Thanks

Thanks

Thank you!

Quite promising.

1. Some of the sources have NEXTGEN diectives. Is full NEXTGEN compatibility on  the horizon?

2. Is the FPC support fully working/stable?

3. Is this meant for FPC with Android target?

4. Smart Mobile Could be an other possible target.

Now I can describe the problem more precisely:
ApplyUpdates works with direct connections, but it does not with remote connections.

Hi,

The last changes in the github trunk are 7 months old. Did this go any further? Has anyone created a real world standalone android application with  mORMot?

Cheers,
Leslie

I think there is a simple way to extend mORMot with the ability to use common tables,   which are much readable for complex queries and  would make recursive queries possible as well.  Should also work with the  NexusDB #temptable format.
Have not tested it yet, but it should only require a little change in the SQLFromSelect function,  plus a few  more method signature needs the extra parameters added with an empty string default value.

NEW PARAMS:

CommonTablesExpr selects the records to work with instead of the whole table, therefor it must have a table with IDs of the records.
IDTableName is the name of this table.

WITH RECURSIVE
  ...
  IDTable(ID) as (...)

function SQLFromSelect(const TableName, Select, Where, SimpleFields, CommonTablesExpr, IDTableName: RawUTF8): RawUTF8;
begin
  if Select='*' then
     // don't send BLOB values to query: retrieve all other fields
    result := 'SELECT '+SimpleFields  else
    result := 'SELECT '+Select;

  // << CODE CHANGE START >>
  if CommonTablesExpr<> ''   then
    result := CommonTablesExpr+' '+result+' FROM '+TableName+' JOIN '+ IDTableName + ' ON '+TableName+'.ID='+IDTableName+'.ID '+SQLFromWhere(Where)
  else
  // << CODE CHANGE END >>
    result := result+' FROM '+TableName+SQLFromWhere(Where);
end;

If having two ID fields in the query is a problem for some db engines it may need an other name in the IDTable (eg _ID, ID_ ... ), or an other parameter for the ID field name, or use the tablename.fieldname syntax.   There is no proper validation for the new params yet.

I need to slow down and look more carefully. It is still not working for me ... will have an other look.

This is what I suspected and  I think this is a serious issue.

Maybe there is a more secure and balanced way to go about this: Instead of versioning the  records it is better to make it possible to identify the Transactions ( or Unit Of Work ... whatever makes the changes atomic). Every record changed by a transaction receives the same Transaction_ID. It makes incremental replication possible. It could go one by one. Or for better performance:  for one replication "transaction"  only as many regular transactions are selected that it can be read and written within a reasonable time. The whole replication may take longer, but ensures atomic behavior during replication and responsiveness  for both master and slaves. It seems reasonable for me to combine it with audit logging. The same history could be used for both. It is much more efficient  to calculate the size and content of the replication transactions in advance  as the change log is being created. When the replication occurs every step is readily available.

I think I have made myself misunderstood. Currently all my test are with  Delphi XE8 (Windows XP & Ubuntu +Wine). That is where TClientDataSet.ApplyUpdates did not work. The rest of the comment was addressing the original question. Luckily  the solution does not require detailed understanding of mORMot, so most of us could probably implement it.  I need to rush with my evaluation of mORMot, so I do ask quite a few questions, while working my way though the documentation, but I would prefer to leave you to the tasks which require your level of understanding of the code, patterns and concepts.  Later on I might be able to implement a few missing things myself.

I have checked this a few days ago with the mORMotVCLTest  sample  and ApplyUpdates did not persist the changes. After refreshing the original values remained.

I the SynDBMidasVCL TClientDataSet is declared as TBufDataset for FPC.   What needed here is to have a TBufDataset descendent implementing  ApplyRecUpdate  to apply the changes via the Mormot Client.

It does not solve the UI binding. TCollection has a private TList<> field. Copying the Class definition and creating a public property to access it may do the trick.

Using TCollection might be a good thing after all. I just need to figure out how to bind it to the UI.

AB,

Thanks for your answers.

1. I have been using Elevate Web Builder for a while. V2.0 has some sort of RTTI. I will check if it is suitable to integrate with mORMot.

2.  ASFAIK Zeoslib droped Kylix support after 7.0. Is it possible to connect to external databases like PG?

3. I am looking for some solution for websockets and call back functionality for Android and IOS. I hope to gain some speed benifit from it as well.
       a) Phonegap has websocket plugin. Is callback interfaces a big step from there? (I might attempt to implement it.)
       b) I will try your bidir socket, but if it does not work out sgcWebsockets client can be an other way to get it done.
       c)  If Websocket is working for FPC client  is  callback interfaces a big step from there?

4. It turned out that the virtual machine needed some more memory to finish linking. The crossplatform client works fine with XE8.

5. I am surprised by this story. I have found the CT people quite polite on their forums.

6. Currently I am testing mORMot compiled with XE8 under linux with wine. The only  issue so far is that sometime the response time can be  up to 5-8 times more with a very simple test case when the  same request is sent to the server in every 500-5000ms. IOCP implementation seems incomplete for wine. Even if it was complete stability is a big question.