DAL Rows and Row from list and dict, respectively

[Lucas]

hello one and all,

isn't this possible?

from pydal.objects import Rows, Row

qFab = Rows([ Row({'word':'happy', 'letters':'fifnen',}), Row({'word':'day', 'letters':'jfinef',}), ]) 

and then:

for f in qFab:

    w, l = f.word, f.letters

    etc....

I seem to remember doing these sorts of things in web2py.

thank you in advance, Lucas

[Alan Etkin]

HI

Have you tried it in a shell?, doesn't seem to be a documented feature

Regards

Alan

[Massimo DiPierro]

I do not think this was every supported. What is the use case?

[Lucas]

ok, to answer both questions.  so, normally, I would db().select() and of course Rows containing Row would return and pass to computations using that data.  but there were times when I generated Rows of Row on-the-fly and temporarily and pass that off to the same computational side.  so the latter is where I would use this on-the-fly Rows of Row to mimic that it came from the db and I wouldn't have to change the computational side.  it's a practical use, believe me.

and so I created a shell script side to test and this is that code:

BOF...

#!/usr/bin/env python
#coding: utf-8
#> xattr -d com.apple.quarantine pydal_row_rows.py
from pydal.objects import Row, Rows

print('opening words.. .')
words = [('the', 26380), ('and', 20367), ('of', 18240), ('to', 15560), ('in', 10501), ('you', 8890), ('a', 7734), ('for', 7691), ('he', 7152), ('lord', 7075), ('i', 6173), ]
print('opened.')

r = Row({'word':words[0][0], 'N':words[0][1]})
print(type(r), r)
print(r.word, r.N, r['word'], r['N'])

# print()
# print('looped.. .')
# qWords1 = Rows()
# for w in words:
# qWords1.append(Row({'word':w[0], 'N':w[1]}))
# print(type(qWords1), len(qWords1))
# for i, w in enumerate(qWords1):
# print(i, w.word, w.N, w['word'], w['N'])
# print('done.')

print()
print('comprehension.. .')
qWords2 = Rows([ Row({'word':w[0], 'N':w[1]}) for w in words ])
print(type(qWords2), len(qWords2))
for i, w in enumerate(qWords2):
print(i, w.word, w.N, w['word'], w['N'])
print('done.')

exit()

EOF.

the most interesting this is that Row seems to work fine in all 3 scenarios.  "comprehension" only works when the "looped" section is uncommented.  definite weird bug in there somewhere.

ok, thank you in advance and sorry for the delay in responding.  Lucas

[Lucas]

perhaps,,, not interesting???

[Massimo DiPierro]

I think this is a bit too much magic. Also what comes from the DB depends on the db and whether or not the data is or is not there may produce different results. I would not recommend doing it. I am surprised it works/worked but I think it should (except perhaps for testing).

[Lucas]

I guess I'm not understanding.  I'm attempting to use the Row and Rows object independent of any database.  just fill and use them strictly from code.  I've got many reasons for doing this.  you did see that the script works when we uncomment the "looped" middle section and doesn't work when it is commented.  try it out and you'll see what I mean.  its quite peculiar.

[Christian Varas]

In my understanding, Rows are not very different from a list containing dictionaries (or Row objects).

Since Rows introduce additional abstraction, internal state, and built-in methods—along with some runtime overhead compared to plain Python lists and dictionaries—what is the main advantage of working with a Rows object instead of using standard data structures?

Is there a significant benefit that makes this approach preferable?

greetings 

--
You received this message because you are subscribed to the Google Groups "py4web" group.
To unsubscribe from this group and stop receiving emails from it, send an email to py4web+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/py4web/f1766ff1-6e78-445a-b257-08873828cb13n%40googlegroups.com.

[Lucas]

not so much with Rows but definitely, yes, with Row.  where Row keys can be addressed using ['key'] as like in dict.  or, like .key as in SimpleNameSpace.  so Row gives the programmer the option of using both interchangeably and that is actually quite nice from a coding point of view.

and since Rows is an extension of Row, it just goes to consistency of use.

[Lucas]

another interesting point.  below is code, updated from above, with both the "looped" and the "comprehension" sections active.  the "looped" loads the list in natural order and the "comprehension" loads the list in the reverse.  but both Rows return their version in the natural order.  that should not be.  coupling that to the first script I submitted, Rows() does not create a new separate and independent instance of the object even though the two instances occupy separate memory spaces.

#!/usr/bin/env python
#coding: utf-8
#> xattr -d com.apple.quarantine pydal_row_rows.py
from pydal.objects import Row, Rows

print('opening words.. .')
words = [('the', 26380), ('and', 20367), ('of', 18240), ('to', 15560), ('in', 10501), ('you', 8890), ('a', 7734), ('for', 7691), ('he', 7152), ('lord', 7075), ('i', 6173), ]
print('opened.')

r = Row({'word':words[0][0], 'N':words[0][1]})
print(type(r), r)
print(r.word, r.N, r['word'], r['N'])

print()
print('looped.. .')
qWords1 = Rows()
for w in words[::+1]:

qWords1.append(Row({'word':w[0], 'N':w[1]}))

print(type(qWords1), len(qWords1), 'addr:', hex(id(qWords1)))

for i, w in enumerate(qWords1):

print(i, w.word, w.N, w['word'], w['N'])
print('done.')

print()
print('comprehension.. .')

qWords2 = Rows([ Row({'word':w[0], 'N':w[1]}) for w in words[::-1] ])
print(type(qWords2), len(qWords2), 'addr:', hex(id(qWords2)))

for i, w in enumerate(qWords2):
print(i, w.word, w.N, w['word'], w['N'])
print('done.')

exit()

[Dave S]

On Tuesday, February 17, 2026 at 12:39:16 PM UTC-8 Lucas wrote:

another interesting point.  below is code, updated from above, with both the "looped" and the "comprehension" sections active.  the "looped" loads the list in natural order and the "comprehension" loads the list in the reverse.  but both Rows return their version in the natural order.  that should not be.  coupling that to the first script I submitted, Rows() does not create a new separate and independent instance of the object even though the two instances occupy separate memory spaces.

FWIW, trying in web2py -S <app>, qWord2 is empty unless qWord1 is defined, and then has the same records.  The address of qWord1[0] and of qWord2[0] match.

Using the comprehension to make an array (foo = []) works as expected.

(Yes, I'm still in "studying mode" with py4web)

/dps

[Lucas]

are you saying my script is wrong?  I don't understand what you're reporting.

but I added the hex(id(qWords1[0])) and hex(id(qWords2[0])) and you're correct that the memory addresses of both are the same and that shouldn't be either, right?  when declaring Row() in each section they should be a new instance in different memory slots.  so what is it with that too?  so weird.

[Dave S]

On Thursday, February 19, 2026 at 5:42:58 AM UTC-8 Lucas wrote:

are you saying my script is wrong?  I don't understand what you're reporting.

I'm reporting that the comprehension by itself creates an empty Rows, and that if the loop is used to create another Rows, either before or after. the C-Rows then has the same contents.

 

but I added the hex(id(qWords1[0])) and hex(id(qWords2[0])) and you're correct that the memory addresses of both are the same and that shouldn't be either, right?  when declaring Row() in each section they should be a new instance in different memory slots.  so what is it with that too?  so weird.

I don't understand it either, but I peeked at the DAL code in web2py, and Rows are more iterators than structures.

/dps