AI Ladder, 0.5

[hughperkins]

hibernate?

https://www.hibernate.org/

It makes storing / retrieving things to/from databases really easy in C# and Java. Basically, you just handle lists and objects in C#/Java, and hibernate handles storing them in the db if they change, and retrieving them from the db when you want it to.

[hughperkins]

To be honest, since in python, data seems to be handed around as dictionaries and lists anyway, it's half done, for the retrieval side, since you can get such a list of dictionaries in two lines:

Code: Select all

dictcursor.execute("select * from maps")
maps = dictcursor.fetchall()

... then the question is how to change some stuff in maps, and inject the stuff that changed back into the db

... or, more simply, and probably 99% of uses, be able to give a single dictionary (=single 'object') to something, and get that to be pumped automatically into a table.

...also, the question of how to handle joins

PS we have headless spring now. Do you fancy setting up a botrunner or two?

[hughperkins]

Perhaps something like:

Code: Select all

def doinsert( tablename, map ):
   global dictcursor

   sqlinsert = "insert into " + tablename + " ("
   isfirstkey = True
   for key in newmap.keys():
      if not isfirstkey:
         sqlinsert = sqlinsert + " , "
      sqlinsert = sqlinsert + key
      isfirstkey = False
   sqlinsert = sqlinsert + ") values ("
   isfirstkey = True
   for value in newmap.values():
      if not isfirstkey:
         sqlinsert = sqlinsert + " , "
      sqlinsert = sqlinsert + " %s "
      isfirstkey = False
   sqlinsert = sqlinsert + ")"
   #print sqlinsert
   dictcursor.execute( sqlinsert, newmap.values() )

Usage:

Code: Select all

newmap = { 'map_name': 'foo1', 'map_archivechecksum': 12343513, 'map_id': 11, 'map_url': '' }
doinsert("maps", newmap )

[hughperkins]

Addendum: we'd also need some way of identifying which fields are the keys so we can choose automatically between inserting and updating.

I don't imagine I'm the first person to want to do this. There's probably something out there somewhere, but I didn't find it yet.

Or maybe I'm looking for the wrong thing, and there's another way of doing this, which is at least as easy, but not quite the same...

[hughperkins]

prototype for simple hibernate for python: http://pastebin.com/f2dfc92da

usage example:

Code: Select all

print pytodb.doselect("maps")

newmap = { 'map_name': 'foo1', 'map_archivechecksum': 12343513, 'map_id': 11, 'map_url': '' }
pytodb.doinsert("maps", newmap )
print pytodb.doselect("maps")

newmap2 = { 'map_name': 'new map name', 'map_archivechecksum': 135135, 'map_id': 11, 'map_url': '' }
pytodb.doinsertorupdate("maps", ['map_id'], newmap2 )
print pytodb.doselect("maps")

pytodb.dodelete( "maps", ['map_id'], newmap )
print pytodb.doselect("maps")

Whether that's easier to use than coding the sql by hand is another question

[imbaczek]

sqlalchemy.

[hughperkins]

sqlalchemy.

Sweet! I'll check that out. The website gives a great first impression, looks professional, so I'm looking forward to checking that out in more detail.

[hughperkins]

Well, that really does seem to behave exactly like hibernate, so I guess what I was looking for was: an object-relational-mapper, and hibernate happens to be one specific implementation of one.

[aegis]

yeah, I used sqlalchemy in uberserver and I'm a fan.

[hughperkins]

yeah, I used sqlalchemy in uberserver and I'm a fan.

Make sure to tell me that *after* I've spent ages figuring it out

[hughperkins]

Release 0.4:

http://manageddreams.com/ailadder/downl ... .0.tar.bz2

Finished migrating to xmlrpc.

[hughperkins]

Right, so sqlalchemy looks like exactly what I was looking for.

There's a new branch sqlalchemy in git where I've started moving to sqlalchemy.

http://github.com/hughperkins/ailadder/tree/sqlalchemy

It seems to work great. Some of the hardest pages are migrated, and look a lot simpler, eg viewrequests.py and viewresults.py.

sqlalchemy needs at least python 2.4 but it also makes it easy to use sqlite and so on for those who want.

[imbaczek]

now you just need to get rid of those fugly prints; use a template language like jinja or mako.

[AF]

Gets a lot further but, error:

error.JPG

(128.16 KiB) Downloaded 11 times

[aegis]

looks like normal internet error

[hughperkins]

looks like normal internet error

So you reckon: just put a try...except around it?

[aegis]

sure.
python has an awesome ability to recover from mostly anything more mild than a segfault, don't waste it

[hughperkins]

now you just need to get rid of those fugly prints; use a template language like jinja or mako.

Ah, sweet! Which do you recommend? Or I just pick one at random and run with that?

I think I will finish migrating to sqlalchemy first though...

You know, it would have been nice to know about this stuff *before* actually coding 80% of the site

[hughperkins]

Gets a lot further but, error:

error.JPG

Hi AF. The verdict seems to be it's a transient error, and the code just needs not to crash during the transient errors.

This new version:

http://manageddreams.com/ailadder/downl ... .1.tar.bz2

... has the getserverrequest surrounded by a try ... except block, so may resolve the issue you are seeing.

[hughperkins]

Guys,

Any hints on the installation process for the website?

I'm thinking, for sqlalchemy, there are a few possibilities:
- ask people to install it themselves, check it is installed
- deploy sqlalchemy with the ailadder setup package, set it up automatically as part of the setup if it's not already installed
- deploy sqlalchemy with the ailadder setup package, and also virtualenv ( http://pypi.python.org/packages/source/ ... 3.4.tar.gz) set it up automatically as part of the setup, to a virtualenv, if it's not already installed
- something else... ?

Thoughts?