Parse uncertainties #1615

MichaelTiemannOSC · 2022-10-16T15:34:11Z

Closes Pint + uncertainties in 2022 #1611, Quantity and Measurement interoperability with uncertainties #1614
Executed pre-commit run --all-files with no errors
The change is fully covered by automated unit tests
Documented in docs/ as appropriate
Added an entry to the CHANGES file

I'd like to discuss and review before committing to "completely cover" all automated test units. I have contributed two test cases that clearly show the filed issues working with the changes. I have run the full test suite, and the one unexpected failure (#783) I cannot explain.

I'm also not quite sure how to approach the docs question. Pint documentation tends to be more on the minimal side, and I don't want to call unnecessary attention to things that should "just work".

Enable Pint to consume uncertain quantities. Signed-off-by: [email protected]

Signed-off-by: MichaelTiemann <[email protected]>

MichaelTiemannOSC · 2022-10-16T16:32:54Z

Related: hgrecco/pint-pandas#140

Signed-off-by: MichaelTiemann <[email protected]>

Signed-off-by: [email protected]

hgrecco · 2022-10-22T19:31:06Z

First, I completely agree with the need of parsing measurements in Pint. I have been looking for this for a while. So thanks, and awesome work.

Having said this, I am not sure about the implementation. Why parsing 8**4 is structurally different from parsing 8+/-4?

In both cases, you have a couple of numbers “connected” by a multichar operator. I guess that the difference is that ** is known by the python tokenizer, while +/- is not.

Take a look at this:

>>> from pint.compat import tokenizer
>>> for p in tokenizer("8**4"):
...  print(p)
...
TokenInfo(type=2 (NUMBER), string='8', start=(1, 0), end=(1, 1), line='8**4')
TokenInfo(type=54 (OP), string='**', start=(1, 1), end=(1, 3), line='8**4')
TokenInfo(type=2 (NUMBER), string='4', start=(1, 3), end=(1, 4), line='8**4')
TokenInfo(type=4 (NEWLINE), string='', start=(1, 4), end=(1, 5), line='')
TokenInfo(type=0 (ENDMARKER), string='', start=(2, 0), end=(2, 0), line='')

So you changed the tokenizer, but in that process it became more difficult to understand, maintain and debug. I wonder if a more sensible approach would not be to add another layer on top of the tokenizer that scans for +/-.

To understand my proposal, let me show you the output of the current tokenizer:

>>> for p in tokenizer("8 + / - 4"):
...  print(p)
...
TokenInfo(type=2 (NUMBER), string='8', start=(1, 0), end=(1, 1), line='8 + / - 4')
TokenInfo(type=54 (OP), string='+', start=(1, 2), end=(1, 3), line='8 + / - 4')
TokenInfo(type=54 (OP), string='/', start=(1, 4), end=(1, 5), line='8 + / - 4')
TokenInfo(type=54 (OP), string='-', start=(1, 6), end=(1, 7), line='8 + / - 4')
TokenInfo(type=2 (NUMBER), string='4', start=(1, 8), end=(1, 9), line='8 + / - 4')
TokenInfo(type=4 (NEWLINE), string='', start=(1, 9), end=(1, 10), line='')
TokenInfo(type=0 (ENDMARKER), string='', start=(2, 0), end=(2, 0), line='')

If we can create function that transform the previous list into:

>>> for p in tokenizer("8 + / - 4"):
...  print(p)
...
TokenInfo(type=2 (NUMBER), string='8', start=(1, 0), end=(1, 1), line='8 + / - 4')
TokenInfo(type=<NEW NUMBER> (OP), string='+ / -', start=(1, 2), end=(1, 7), line='8 + / - 4')
TokenInfo(type=2 (NUMBER), string='4', start=(1, 8), end=(1, 9), line='8 + / - 4')
TokenInfo(type=4 (NEWLINE), string='', start=(1, 9), end=(1, 10), line='')
TokenInfo(type=0 (ENDMARKER), string='', start=(2, 0), end=(2, 0), line='')

(this could also handle the parenthesis)

Then the new tokenizer will just be:

def tokenizer(s):
    return uncertainty_layer(old_tokenizer(s))

(the names are crap, but just as an example) So it is easy to debug and turn on/off if necessary.

If you go this way, I think it will be wise to move the tokenizer function out of the compat module and into pint_eval.

hgrecco · 2022-10-22T20:09:45Z

I also think we need to clearly define which uncertainties format we are going to support (and which we are not).

MichaelTiemannOSC · 2022-10-22T20:18:42Z

Thank you so much for giving it a look.

On the choice of creating a tighter binding than **, consider `x ** 2+/-0 ** (1/2)'. If +/- binds the same as **, it will fail when there's a string of tokens that all have the same precedence (but different associativity).

I don't disagree with you at all that there are better ways to do this. What I wanted to do first was a proof-of-concept, to see whether the bones in Pint and Pint-Pandas were ready for this sort of thing. And I think they are. My principle concern in modifying the tokenizer was out of respect for the fact that Pint does talk to itself using str() and repr, and as such we need to be prepared to handle whatever sorts of formats users may throw at us (including the N(S) notation and paranthesized ± as a unicode symbol).

Please keep the thoughts coming!

hgrecco · 2022-10-22T21:24:57Z

On the choice of creating a tighter binding than **, consider `x ** 2+/-0 ** (1/2)'. If +/- binds the same as **, it will fail when there's a string of tokens that all have the same precedence (but different associativity).

I am not sure that I am proposing the same thing. Consider parsing 8 + / - 4:

TokenInfo(type=2 (NUMBER), string='8', start=(1, 0), end=(1, 1), line='8 + / - 4')
TokenInfo(type=54 (OP), string='+', start=(1, 2), end=(1, 3), line='8 + / - 4')
TokenInfo(type=54 (OP), string='/', start=(1, 4), end=(1, 5), line='8 + / - 4')
TokenInfo(type=54 (OP), string='-', start=(1, 6), end=(1, 7), line='8 + / - 4')
TokenInfo(type=2 (NUMBER), string='4', start=(1, 8), end=(1, 9), line='8 + / - 4')

which is transformed to:

TokenInfo(type=2 (NUMBER), string='8', start=(1, 0), end=(1, 1), line='8 + / - 4')
TokenInfo(type=<NEW NUMBER> (OP), string='+ / -', start=(1, 2), end=(1, 7), line='8 + / - 4')
TokenInfo(type=2 (NUMBER), string='4', start=(1, 8), end=(1, 9), line='8 + / - 4')

which is then converted by build_eval_tree to:

ufloat_fromstr("8 +/- 4")

This will automatically support every format supported here. I think we should limit to those (at least for the time being).

Consider now parsing (8 + / - 4) m:

TokenInfo(type=54 (OP), string='(', start=(1, 0), end=(1, 1), line='(8 +/- 4) m')
TokenInfo(type=2 (NUMBER), string='8', start=(1, 1), end=(1, 2), line='(8 +/- 4) m')
TokenInfo(type=54 (OP), string='+', start=(1, 3), end=(1, 4), line='(8 +/- 4) m')
TokenInfo(type=54 (OP), string='/', start=(1, 4), end=(1, 5), line='(8 +/- 4) m')
TokenInfo(type=54 (OP), string='-', start=(1, 5), end=(1, 6), line='(8 +/- 4) m')
TokenInfo(type=2 (NUMBER), string='4', start=(1, 7), end=(1, 8), line='(8 +/- 4) m')
TokenInfo(type=54 (OP), string=')', start=(1, 8), end=(1, 9), line='(8 +/- 4) m')
TokenInfo(type=1 (NAME), string='m', start=(1, 10), end=(1, 11), line='(8 +/- 4) m')

which will then be converted

TokenInfo(type=54 (OP), string='(', start=(1, 0), end=(1, 1), line='(8 +/- 4) m')
TokenInfo(type=2 (NUMBER), string='8', start=(1, 1), end=(1, 2), line='(8 +/- 4) m')
TokenInfo(type=<NEW NUMBER> (OP), string='+ / -', start=(1, 3), end=(1, 6), line='(8 +/- 4) m')
TokenInfo(type=2 (NUMBER), string='4', start=(1, 7), end=(1, 8), line='(8 +/- 4) m')
TokenInfo(type=54 (OP), string=')', start=(1, 8), end=(1, 9), line='(8 +/- 4) m')
TokenInfo(type=1 (NAME), string='m', start=(1, 10), end=(1, 11), line='(8 +/- 4) m')

which is then converted by build_eval_tree to:

ufloat_fromstr("8 +/- 4") * ureg.meter

(I am 'paraphrasing' here)

Please note that that Pint supports more numerical types than uncertainties. We need to make this clear in the docs

MichaelTiemannOSC · 2022-11-17T01:33:33Z

I have returned from a long trip and am ready to tackle this now. I like the proposed approach (of creating a layer that can be enabled and disabled) and will give it a go. However, this idea--as I understand it--is a no go, because OP is (for better or worse), number 54:

TokenInfo(type=<NEW NUMBER> (OP), string='+ / -', start=(1, 3), end=(1, 6), line='(8 +/- 4) m')

To give this string a new number would mean it's no longer an OP, but something entirely different, which would mean we'd need to special-case its handling as something-other-than-OP. Would it be your preference to insert a new kind of OP (such as NUMBER_CONSTRUCTOR)? Or, given that OP is OP is OP, leave that alone and focus only on implementing a pre-processor as a general function and then parsing and evaluating the results of the preprocessor separately?

But also, before I go too far, I have to point out that a very common format to get back from Pint (when dealing with large quantities) is more like this:

(8.0 + / - 4.0)e6 t CO2

This creates the problem of deciding when and how to distribute the factor of a million (e6) so as to end up with the result:

ufloat(8.0e6, 4.0e6) t CO2

We cannot simply process what's between the (); rather there's some additional look-ahead and processing necessary. So I'm looking at that.

I should add that I'm a rank notice when it comes to pytest and friends. My plan is to test just tokenization and basic processing of uncertain quantities within the context of test_measurement.py

I will leave it to others to flip the magic switches so that test_umath and friends can, if they want to, test the actual mechanics of uncertainties within Pint. Hope that's OK!

Based on feedback, tokenize uncertainties on top of default tokenizer, not instead of default tokenizer. Signed-off-by: MichaelTiemann <[email protected]>

MichaelTiemannOSC · 2022-11-18T03:48:22Z

The original pint tokenizer is _plain_tokenizer and the tokenizer that layers on top to handle uncertainties is uncertainty_tokenizer. I've created a tokenizer variable so that users can switch away from the default this way:

from pint import pint_eval
pint_eval.tokenizer = pint_eval.uncertainty_tokenizer

I'm very open to changing names and/or mechanics of how the tokenizer should be selected. I've not yet started to flesh out some better test cases for test_measurement.py. I'd like to get feedback on these changes first.

MichaelTiemannOSC · 2022-11-18T07:40:40Z

pint/pint_eval.py

 from .errors import DefinitionSyntaxError

 # For controlling order of operations
 _OP_PRIORITY = {
+    "±": 4,


This line can go away: we change ± to +/- to avoid gratuitous problems with ERRORTOKEN.

MichaelTiemannOSC · 2022-11-28T16:29:25Z

Any comments on this PR? I think I made the changes that were requested.

MichaelTiemannOSC · 2022-12-13T15:17:21Z

I'm still waiting, and my project is still waiting to know if this is ready to enter final approach. We'd really like to add uncertainties to our pint-based work, but are loathe to ask users to install a completely custom version of pint. If there are other changes you would like to see, I would be happy to receive feedback!

MichaelTiemannOSC · 2022-12-28T23:55:35Z

pint/pint_eval.py

 _BINARY_OPERATOR_MAP = {
+    "±": _ufloat,


The line "±": _ufloat, can probably also go away, as above.

MichaelTiemannOSC · 2022-12-28T23:56:10Z

pint/pint_eval.py

+        if tokinfo.type != tokenlib.ENCODING:
+            yield tokinfo
+
+def uncertainty_tokenizer(input_string):


Would uncertainties_tokenizer be a better name?

Signed-off-by: Michael Tiemann <[email protected]>

Fixes problems parsing currency symbols that also show up when dealing with uncertainties. Signed-off-by: Michael Tiemann <[email protected]>

Handle negative numbers using uncertainty parenthesis notation. Signed-off-by: Michael Tiemann <[email protected]>

Ahem...use walrus operator for side-effect, not truth value. Signed-off-by: Michael Tiemann <[email protected]>

… of the exponent, not just in the parsing of the exponent. i.e., (5.01+/-0.07)e+04 Signed-off-by: Michael Tiemann <[email protected]>

MichaelTiemannOSC · 2023-02-22T08:59:15Z

Can somebody please review?

jakeogh · 2023-04-16T21:27:16Z

While trying to understand how to fix #1646 I pulled this branch and ran the tests:

Before parse-uncertainties:

platform linux -- Python 3.10.11, pytest-7.3.0, pluggy-1.0.0                                                                                                                                                                                                                     
rootdir: /home/cfg/_myapps/pint                                                                                                                                                                                                                                                  
plugins: typeguard-3.0.2, pkgcore-0.12.20, anyio-3.6.1                                                                                                                                                                                                                           
collected 2080 items / 2 skipped
<snip>
2 failed, 2024 passed, 13 skipped, 11 xfailed, 11 warnings, 32 errors

After parse-uncertainties:

platform linux -- Python 3.10.11, pytest-7.3.0, pluggy-1.0.0                                                                                                                                                                                                                     
rootdir: /home/cfg/_myapps/pint                                                                                                                                                                                                                                                  
plugins: typeguard-3.0.2, pkgcore-0.12.20, anyio-3.6.1                                                                                                                                                                                                                           
collected 2083 items / 2 skipped
<snip>
11 failed, 2018 passed, 13 skipped, 11 xfailed, 13 warnings, 32 errors

pint-convert is unfortunately still broken (I rebased, so no surprise):

$ pint-convert 1ft meter
Traceback (most recent call last):
  File "/usr/lib/python-exec/python3.10/pint-convert", line 109, in <module>
    ureg._units["R_inf"].converter.scale = R_i
  File "<string>", line 4, in __setattr__
dataclasses.FrozenInstanceError: cannot assign to field 'scale'

In parallel with Pint and Pint-Pandas, changes to support uncertainties as an EA Dtype. See hgrecco/pint#1615 and hgrecco/pint-pandas#140 Signed-off-by: Michael Tiemann <[email protected]>

With these changes (and pandas-dev/pandas#53970 and hgrecco/pint#1615) the test suite passes or xpasses everything (no failures or error). Indeed, the uncertainties code has essentially doubled the scope of the test suite (to test with and without it). The biggest gotcha is that the EA for complex numbers is not compatible with the EA for uncertainties, due to incompatible hacks: The hack for complex numbers is to np.nan (which is, technically, a complex number) for na_value across all numeric types. But that doesn't work for uncertainties, because uncertainties doesn't accept np.nan as an uncertain value. The hack for uncertainties is to use pd.NA for na_value. This works for Int64, Float64, and uncertainties, but doesn't work for complex (which cannot tolerate NAType). Some careful subclassing fills in what doesn't easily work, with fixtures to prevent the improper mixing of complex and uncertainty types in the same python environment. Happy to discuss! Signed-off-by: Michael Tiemann <[email protected]>

MichaelTiemannOSC · 2023-09-14T16:35:11Z

Shall we merge?

hgrecco · 2023-09-14T20:39:10Z

Yes but Any idea why the test suite is not running for this PR?

MichaelTiemannOSC · 2023-09-15T10:40:48Z

I don't know how the CI/CD works. I certainly haven't tried to do anything to evade it. Does anybody else know?

keewis · 2023-09-15T10:53:16Z

as far as I can tell, for PRs of new contributors (as far as the commit history is concerned) the maintainers have to manually confirm every time there's commits.

This is configurable (we have switched it off in xarray), but I think github's intention is to by default avoid running CI with potentially malicious changes (like, cryptocurrency) before having a maintainer review those changes.

keewis · 2023-09-15T10:58:29Z

it is strange, though, that it doesn't tell you that. Maybe try closing / reopening the PR?

MichaelTiemannOSC · 2023-09-15T11:05:07Z

Closing PR to reopen it (attempting soft reboot of CI/CD).

MichaelTiemannOSC · 2023-09-15T11:05:42Z

Reopening PR to attempt soft reboot of CI/CD.

MichaelTiemannOSC · 2023-09-15T11:08:25Z

It looks like there's a problem with a readthedocs key on Pint's side.

keewis · 2023-09-15T11:31:21Z

no, it's just that the configuration (.readthedocs.yaml) needs an update. As far as I can tell, system_packages: false can be dropped without any change in behavior. This would be an easy fix if you want to send in a PR.

After that is merged, you also wouldn't have to wait on maintainer approval for CI to run on this PR.

hgrecco · 2023-09-15T11:53:42Z

Now it seems to be running. If all test pass, I will merge right away!

codspeed-hq · 2023-09-15T12:11:12Z

CodSpeed Performance Report

Merging #1615 will not alter performance

_{Comparing MichaelTiemannOSC:parse-uncertainties (00f08f3) with master (2852f36)}

Summary

✅ 421 untouched benchmarks

hgrecco · 2023-09-15T13:03:02Z

Test are failing. There seems to be an error due to missing babel dependency for a test.

Here's hoping this fixes the CI/CD problem with test_1400. Signed-off-by: Michael Tiemann <[email protected]>

MichaelTiemannOSC · 2023-09-15T16:55:19Z

I'm taking a wild guess that because we now have a test case that depends on babel, we need babel as part of testbase. So I'm adding that there and saying a small prayer. If somebody knows how to actually fix this, I'm all ears!

@keewis

Removing `system_packages: false` as suggested by @keewis Signed-off-by: Michael Tiemann <[email protected]>

MichaelTiemannOSC · 2023-09-15T17:06:09Z

Not my day: lint failed for a reason I don't understand/cannot reproduce: reserveCache failed: Cache already exists. Scope: refs/heads/parse-uncertainties, Key: [xxx], Version: 25d34710bdd63bed70a22b8204c3e1d0942dc237e098600e3896b4334c6d784f

hgrecco · 2023-09-15T17:08:40Z

It is better that you use the decorator that is provided for this purpose:

pint/pint/testsuite/helpers.py

Line 125 in 2852f36

requires_babel = pytest.mark.skipif(

Just import requires_babel and add @requires_babel to any test that requires babel.

Do not add babel as a requirement as it will make all tests slower.

MichaelTiemannOSC · 2023-09-15T17:11:47Z

Thanks, I'll fix that. There are some new failures I need to fix, which I will:

FAILED pint/testsuite/test_compat.py::test_isnan[False] - TypeError: must be real number, not datetime.datetime
FAILED pint/testsuite/test_compat.py::test_isnan[True] - TypeError: must be real number, not datetime.datetime
FAILED pint/testsuite/test_compat.py::test_zero_or_nan[False] - TypeError: must be real number, not datetime.datetime
FAILED pint/testsuite/test_compat.py::test_zero_or_nan[True] - TypeError: must be real number, not datetime.datetime

Fix isnan to use unp.isnan as appropriate for both duck_array_type and objects of UFloat types. Fix a minor typo in pint/facets/__init__.py comment. In test_issue_1400, use decorators to ensure babel library is loaded when needed. pyproject.toml: revert change to testbase; we fixed with decorators instead. Signed-off-by: Michael Tiemann <[email protected]>

hgrecco · 2023-09-15T19:40:41Z

Looks good! Shall we merge @MichaelTiemannOSC ?

MichaelTiemannOSC · 2023-09-15T19:51:17Z

Yes, thanks 🙏

hgrecco · 2023-09-15T19:59:00Z

Congratulations and thanks for this excelent work!!

MichaelTiemannOSC added 3 commits October 13, 2022 21:14

Parse uncertain numbers e.g. (1.0+/-0.2)e+03

3e743a3

Enable Pint to consume uncertain quantities. Signed-off-by: [email protected]

Fix problems identified by python -m pre_commit run --all-files

a54c597

Signed-off-by: MichaelTiemann <[email protected]>

Enhance support for uncertainties. See hgrecco#1611, hgrecco#1614.

7d2fada

Signed-off-by: MichaelTiemann <[email protected]>

MichaelTiemannOSC mentioned this pull request Oct 17, 2022

Unexpected results for the mean of a DataFrame of ufloat from the uncertainties package. pandas-dev/pandas#14162

Closed

MichaelTiemannOSC added 2 commits October 19, 2022 08:00

Fix up failures and errors found by test suite.

fc8564b

Signed-off-by: MichaelTiemann <[email protected]>

Copy in changes from PR1596

c8fe27f

Signed-off-by: [email protected]

MichaelTiemannOSC mentioned this pull request Oct 21, 2022

Adopt NEP-37 and NEP-47 (array module and Python Array API) #1592

Open

Create modular uncertainty parser layer

126a859

Based on feedback, tokenize uncertainties on top of default tokenizer, not instead of default tokenizer. Signed-off-by: MichaelTiemann <[email protected]>

MichaelTiemannOSC commented Nov 18, 2022

View reviewed changes

jules-ch mentioned this pull request Nov 29, 2022

String parsing to initiate Measurement object #1672

Closed

MichaelTiemannOSC commented Dec 28, 2022

View reviewed changes

MichaelTiemannOSC added 6 commits December 29, 2022 14:59

Resolving conflicts against pint 20.1

d17b70c

Signed-off-by: Michael Tiemann <[email protected]>

Fix conflict merge error

f89e183

Signed-off-by: Michael Tiemann <[email protected]>

Update util.py

7198cf0

Fixes problems parsing currency symbols that also show up when dealing with uncertainties. Signed-off-by: Michael Tiemann <[email protected]>

Update pint_eval.py

e5004a5

Handle negative numbers using uncertainty parenthesis notation. Signed-off-by: Michael Tiemann <[email protected]>

Update pint_eval.py

a4a1fa5

Ahem...use walrus operator for side-effect, not truth value. Signed-off-by: Michael Tiemann <[email protected]>

Fixed to work with both + and - e notation in the actually processing…

7938056

… of the exponent, not just in the parsing of the exponent. i.e., (5.01+/-0.07)e+04 Signed-off-by: Michael Tiemann <[email protected]>

MichaelTiemannOSC mentioned this pull request Feb 22, 2023

Investigate integrating SEC 10-K EX-21 subsidiary owner info with PUDL catalyst-cooperative/pudl#2225

Closed

MichaelTiemannOSC mentioned this pull request Jul 2, 2023

Support uncertainties as EA Dtype pandas-dev/pandas#53970

Closed

5 tasks

MichaelTiemannOSC mentioned this pull request Jul 5, 2023

Make AffineScalarFunc hashable for Pandas, Pint and Pint-Pandas lmfit/uncertainties#170

Closed

Merge remote-tracking branch 'origin/master' into parse-uncertainties

10e07ea

MichaelTiemannOSC closed this Sep 15, 2023

MichaelTiemannOSC reopened this Sep 15, 2023

Make babel a dependency for testbase

4e20d99

Here's hoping this fixes the CI/CD problem with test_1400. Signed-off-by: Michael Tiemann <[email protected]>

Update .readthedocs.yaml

f55b8de

Removing `system_packages: false` as suggested by @keewis Signed-off-by: Michael Tiemann <[email protected]>

hgrecco merged commit 07646d0 into hgrecco:master Sep 15, 2023
33 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Parse uncertainties #1615

Parse uncertainties #1615

MichaelTiemannOSC commented Oct 16, 2022

MichaelTiemannOSC commented Oct 16, 2022

hgrecco commented Oct 22, 2022

hgrecco commented Oct 22, 2022

MichaelTiemannOSC commented Oct 22, 2022

hgrecco commented Oct 22, 2022

MichaelTiemannOSC commented Nov 17, 2022 •

edited

Loading

MichaelTiemannOSC commented Nov 18, 2022

MichaelTiemannOSC Nov 18, 2022

MichaelTiemannOSC commented Nov 28, 2022

MichaelTiemannOSC commented Dec 13, 2022

MichaelTiemannOSC Dec 28, 2022

MichaelTiemannOSC Dec 28, 2022

MichaelTiemannOSC commented Feb 22, 2023

jakeogh commented Apr 16, 2023 •

edited

Loading

MichaelTiemannOSC commented Sep 14, 2023

hgrecco commented Sep 14, 2023

MichaelTiemannOSC commented Sep 15, 2023

keewis commented Sep 15, 2023 •

edited

Loading

keewis commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

keewis commented Sep 15, 2023 •

edited

Loading

hgrecco commented Sep 15, 2023

codspeed-hq bot commented Sep 15, 2023 •

edited

Loading

hgrecco commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

hgrecco commented Sep 15, 2023 •

edited

Loading

MichaelTiemannOSC commented Sep 15, 2023

hgrecco commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

hgrecco commented Sep 15, 2023

Parse uncertainties #1615

Parse uncertainties #1615

Conversation

MichaelTiemannOSC commented Oct 16, 2022

MichaelTiemannOSC commented Oct 16, 2022

hgrecco commented Oct 22, 2022

hgrecco commented Oct 22, 2022

MichaelTiemannOSC commented Oct 22, 2022

hgrecco commented Oct 22, 2022

MichaelTiemannOSC commented Nov 17, 2022 • edited Loading

MichaelTiemannOSC commented Nov 18, 2022

MichaelTiemannOSC Nov 18, 2022

Choose a reason for hiding this comment

MichaelTiemannOSC commented Nov 28, 2022

MichaelTiemannOSC commented Dec 13, 2022

MichaelTiemannOSC Dec 28, 2022

Choose a reason for hiding this comment

MichaelTiemannOSC Dec 28, 2022

Choose a reason for hiding this comment

MichaelTiemannOSC commented Feb 22, 2023

jakeogh commented Apr 16, 2023 • edited Loading

MichaelTiemannOSC commented Sep 14, 2023

hgrecco commented Sep 14, 2023

MichaelTiemannOSC commented Sep 15, 2023

keewis commented Sep 15, 2023 • edited Loading

keewis commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

keewis commented Sep 15, 2023 • edited Loading

hgrecco commented Sep 15, 2023

codspeed-hq bot commented Sep 15, 2023 • edited Loading

CodSpeed Performance Report

Merging #1615 will not alter performance

Summary

hgrecco commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

hgrecco commented Sep 15, 2023 • edited Loading

MichaelTiemannOSC commented Sep 15, 2023

hgrecco commented Sep 15, 2023

MichaelTiemannOSC commented Sep 15, 2023

hgrecco commented Sep 15, 2023

MichaelTiemannOSC commented Nov 17, 2022 •

edited

Loading

jakeogh commented Apr 16, 2023 •

edited

Loading

keewis commented Sep 15, 2023 •

edited

Loading

keewis commented Sep 15, 2023 •

edited

Loading

codspeed-hq bot commented Sep 15, 2023 •

edited

Loading

hgrecco commented Sep 15, 2023 •

edited

Loading