Implement f-string support #3250
Comments
marcino239
changed the title
|
Thanks for the report. The |
|
If you don't mid asking, what is the error message trying to mention. I got similar sort of error but don't know how to understand the error message |
|
For implementing the |
|
@rtmatx the error message is mentioning that the opcode |
|
@anthonytec2 The |
|
#3938 implements a better error message in such cases. For example, in the case above it will print: |
|
hi guys, any workaround? |
|
@hbaaron unfortunately there is no work around, use of format string (and hence that opcode) is not yet supported. |
|
I have been using formatted strings ( |
…dError from numba NotImplementedError: Failed in object mode pipeline (step: analyzing bytecode) Use of unknown opcode FORMAT_VALUE see numba/numba#3250 #306
|
@monocongo I'm not sure what's going on in your case, In [1]: def foo(values):
...: return f"string {values}"
...:
In [2]: def bar(values):
...: return "string {}".format(values)
...:
In [3]: from dis import dis
In [4]: dis(foo)
2 0 LOAD_CONST 1 ('string ')
2 LOAD_FAST 0 (values)
4 FORMAT_VALUE 0
6 BUILD_STRING 2
8 RETURN_VALUE
In [5]: from numba import jit
In [6]: jit(foo)(10)
---------------------------------------------------------------------------
UnsupportedError: Failed in object mode pipeline (step: analyzing bytecode)
Use of unknown opcode 'FORMAT_VALUE'
File "<ipython-input-1-2130adc6315a>", line 2:
def foo(values):
return f"string {values}"
^
Unsupported functionality was found in the code Numba was trying to compile.
If this functionality is important to you please file a feature request at:
https://github.com/numba/numba/issues/new
In [7]: jit(bar)(10)
Out[7]: 'string 10'
In [8]: jit(nopython=True)(bar)(10)
TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Unknown attribute 'format' of type Literal[str](string {})
File "<ipython-input-2-852f47965480>", line 2:
def bar(values):
return "string {}".format(values)
^
[1] During: typing of get attribute at <ipython-input-2-852f47965480> (2)
File "<ipython-input-2-852f47965480>", line 2:
def bar(values):
return "string {}".format(values)
^
In [9]: dis(bar)
2 0 LOAD_CONST 1 ('string {}')
2 LOAD_METHOD 0 (format)
4 LOAD_FAST 0 (values)
6 CALL_METHOD 1
8 RETURN_VALUEEven with a dead branch, the op code would still need to be interpreted... In [10]: def foo_branch(values, jmp):
...: if jmp > 0:
...: return f"string {values}"
...: else:
...: return "const str"
...:
...:
In [11]: dis(foo_branch)
2 0 LOAD_FAST 1 (jmp)
2 LOAD_CONST 1 (0)
4 COMPARE_OP 4 (>)
6 POP_JUMP_IF_FALSE 18
3 8 LOAD_CONST 2 ('string ')
10 LOAD_FAST 0 (values)
12 FORMAT_VALUE 0
14 BUILD_STRING 2
16 RETURN_VALUE
5 >> 18 LOAD_CONST 3 ('const str')
20 RETURN_VALUE
22 LOAD_CONST 0 (None)
24 RETURN_VALUE
In [12]: jit(foo_branch)(10, -1)
UnsupportedError: Failed in object mode pipeline (step: analyzing bytecode)
Use of unknown opcode 'FORMAT_VALUE'
File "<ipython-input-10-c5ac4bf10bb6>", line 3:
def foo_branch(values, jmp):
<source elided>
if jmp > 0:
return f"string {values}"
^Also, |
to work around the numba issue described here: numba/numba#3250
|
Since this is a feature request and it is probably still wanted, I'll change the title now to reflect what is being requested. |
esc
changed the title
|
Thanks, @esc. It'll be nice to be able to use f-strings in numba-annotated code once this is ready to go. I appreciate the useful work you guys do on numba, thanks! |
|
@monocongo thanks for the kind words! We appreciate you using Numba. 😄 |
|
still no support? |
|
@adizhol that is correct, so sorry. |
|
Can I use print with argument 'end' like this: print('something', end='\r')? |
|
Thanks for making numba, it is incredibly cool and powerful! I am using numba as I would like to use f-strings but I get the same numba exception as described above. Would it be possible to add support for f-strings when numba is running in forced object mode? Thanks! |
|
xref #6608, which provides basic support. |
|
please fix this |
|
5 years later, any updates? |
|
Unfortunately, I also get this error. |
|
This is clearly looking for someone to volunteer to implement it. Please don't hesitate to consider this if you need it for your Numba use-case. |
This program:
Throws this exception:
The text was updated successfully, but these errors were encountered: