use the formula environment as the newly created quosure environment.

[romainfrancois]

closes #5881

# pak::pak("msberends/AMR@v1.6.0")
library(AMR)
library(skimr)

skim(example_isolates)

Name	example_isolates
Number of rows	2000
Number of columns	49
_______________________
Column type frequency:
character	2
Date	1
factor	1
logical	3
mo	1
numeric	1
rsi	40
________________________
Group variables	None

Data summary

Variable type: character

skim_variable	n_missing	complete_rate	min	max	empty	n_unique	whitespace
gender	0	1	1	1	0	2	0
patient_id	0	1	6	6	0	981	0

Variable type: Date

skim_variable	n_missing	complete_rate	min	max	median	n_unique
date	0	1	2002-01-02	2017-12-28	2009-07-31	1140

Variable type: factor

skim_variable	n_missing	complete_rate	ordered	n_unique	top_counts
hospital_id	0	1	FALSE	4	D: 762, B: 663, A: 321, C: 254

Variable type: logical

skim_variable	n_missing	complete_rate	mean	count
ward_icu	0	1	0.32	FAL: 1354, TRU: 646
ward_clinical	0	1	0.73	TRU: 1464, FAL: 536
ward_outpatient	0	1	0.06	FAL: 1880, TRU: 120

Variable type: mo

skim_variable	n_missing	complete_rate	unique_total	gram_negative	gram_positive	top_genus	top_species
mo	0	1	90	NA	NA	Staphylococcus	Escherichia coli

Variable type: numeric

skim_variable	n_missing	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
age	0	1	71.06	14.12	14	63	74	82	97	▁▁▃▇▅

Variable type: rsi

skim_variable	n_missing	complete_rate	ab_name	count_R	count_S	count_I	prop_R	prop_S	prop_I
PEN	371	0.81	NA	1201	428	11	0.74	0.26	0.01
OXA	1635	0.18	NA	114	251	0	0.31	0.69	0.00
FLC	1057	0.47	NA	278	665	0	0.29	0.71	0.00
AMX	650	0.68	NA	804	546	3	0.60	0.40	0.00
AMC	121	0.94	NA	446	1433	91	0.24	0.76	0.05
AMP	650	0.68	NA	804	546	3	0.60	0.40	0.00
TZP	999	0.50	NA	126	875	13	0.13	0.87	0.01
CZO	1554	0.22	NA	199	247	2	0.45	0.55	0.00
FEP	1276	0.36	NA	103	621	1	0.14	0.86	0.00
CXM	211	0.89	NA	470	1319	22	0.26	0.74	0.01
FOX	1182	0.41	NA	224	594	8	0.27	0.73	0.01
CTX	1057	0.47	NA	146	797	1	0.15	0.85	0.00
CAZ	189	0.91	NA	1204	607	0	0.66	0.34	0.00
CRO	1057	0.47	NA	146	797	1	0.15	0.85	0.00
GEN	145	0.93	NA	456	1399	27	0.25	0.75	0.01
TOB	649	0.68	NA	465	886	7	0.34	0.66	0.01
AMK	1308	0.35	NA	441	251	0	0.64	0.36	0.00
KAN	1529	0.24	NA	471	0	0	1.00	0.00	0.00
TMP	501	0.75	NA	571	928	10	0.38	0.62	0.01
SXT	241	0.88	NA	361	1398	6	0.21	0.79	0.00
NIT	1257	0.37	NA	127	616	51	0.17	0.83	0.07
FOS	1649	0.18	NA	148	203	0	0.42	0.58	0.00
LNZ	977	0.51	NA	709	314	0	0.69	0.31	0.00
CIP	591	0.70	NA	228	1181	69	0.16	0.84	0.05
MFX	1789	0.11	NA	71	140	4	0.34	0.66	0.02
VAN	139	0.93	NA	712	1149	0	0.38	0.62	0.00
TEC	1024	0.49	NA	739	237	0	0.76	0.24	0.00
TCY	800	0.60	NA	357	843	23	0.30	0.70	0.02
TGC	1202	0.40	NA	101	697	0	0.13	0.87	0.00
DOX	864	0.57	NA	315	821	7	0.28	0.72	0.01
ERY	106	0.95	NA	1084	810	9	0.57	0.43	0.00
CLI	480	0.76	NA	930	590	4	0.61	0.39	0.00
AZM	106	0.95	NA	1084	810	9	0.57	0.43	0.00
IPM	1111	0.44	NA	55	834	10	0.06	0.94	0.01
MEM	1171	0.41	NA	49	780	0	0.06	0.94	0.00
MTR	1966	0.02	NA	5	29	0	0.15	0.85	0.00
CHL	1846	0.08	NA	33	121	0	0.21	0.79	0.00
COL	360	0.82	NA	1331	309	0	0.81	0.19	0.00
MUP	1730	0.14	NA	16	254	3	0.06	0.94	0.01
RIF	997	0.50	NA	698	305	2	0.70	0.30	0.00

^{Created on 2021-05-17 by the reprex package (v2.0.0)}

[lionel-]

I was planning for R 4.1 lambdas and trying to have formulas and native lambdas use the same code path. However I think that's a mistake, we should have different code paths.

• Evaluated formulas can be inlined.
• Evaluated functions cannot be inlined.

So there will be performance implications for using list(~ f(.), ~ g(.)) over list(\(.) f(.), \. g(.)). The latter won't be inlined. Maybe we will be able to do better with strictness annotations.

The bottom line is that I think your change makes sense and we should accept the different semantics for r-lib and native lambdas. I have added a unit test to cover the local lookup case.