update ref_group column ordering

[clarkliming]

close #747

[github-actions]

Code Coverage Summary

Filename                     Stmts    Miss  Cover    Missing
-------------------------  -------  ------  -------  ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
R/00tabletrees.R               737      62  91.59%   22, 103, 106, 413, 497-498, 501, 656, 756, 847-848, 949, 951-952, 975-978, 1000, 1113-1116, 1211-1216, 1363, 1458-1461, 1524-1527, 1562-1565, 1571-1576, 1625, 1632, 1730, 1840, 1854, 1857-1860, 1863-1866, 1894, 1925-1926
R/as_html.R                    138       6  95.65%   66, 133-137
R/colby_constructors.R         555      20  96.40%   71, 123, 181-184, 244-247, 387, 404, 1213, 1304, 1465, 1503, 1525, 1549, 1570, 1726
R/compare_rtables.R             83      17  79.52%   93-96, 99-102, 115-118, 137, 157-158, 189, 194
R/format_rcell.R                12       0  100.00%
R/indent.R                      13       2  84.62%   39-40
R/index_footnotes.R             62       0  100.00%
R/make_split_fun.R             119      23  80.67%   22-25, 51-54, 57-60, 115, 119, 280, 283-286, 291-294, 313, 412
R/make_subset_expr.R           136      14  89.71%   34-48, 126-133, 168, 250, 266, 274
R/simple_analysis.R              5       1  80.00%   55
R/split_funs.R                 505      66  86.93%   143, 148, 154-159, 164, 181-185, 366-371, 388-393, 474, 526, 544-547, 564, 631, 641-642, 644, 658, 702, 727, 903, 910, 936-939, 950-951, 953, 955, 1126-1128, 1142-1146, 1210-1213, 1276-1279
R/summary.R                    215      24  88.84%   38, 85, 192, 200, 271-276, 287-288, 307-308, 418, 465-481, 516, 549
R/tree_accessors.R             833      79  90.52%   109, 251, 269, 292, 330, 344, 360, 465, 492-493, 774-779, 907, 925, 949, 999, 1054-1055, 1094, 1127, 1163-1167, 1220, 1295-1299, 1317-1327, 1396, 1501-1504, 1529, 1549-1550, 1559, 1600, 1618-1622, 1643-1647, 1726, 1768, 1872, 1976, 1989, 2002, 2016, 2024, 2033-2037, 2379, 2737, 2850, 2944-2951, 3111, 3229, 3255-3289, 3305-3312
R/tt_afun_utils.R              411      32  92.21%   50, 164, 171, 181-194, 260, 271-272, 504, 512-515, 597-601, 622, 637-639
R/tt_compare_tables.R           70       4  94.29%   56, 178, 257, 261
R/tt_compatibility.R           510      56  89.02%   19, 142-143, 192, 197, 332-333, 337-340, 346, 350, 398, 520, 568, 601, 621, 654-657, 701, 718-722, 809, 837-840, 849, 912, 920, 931-934, 1044, 1051, 1080-1094, 1125-1126
R/tt_dotabulation.R           1116      96  91.40%   54, 252, 257, 259, 310, 334, 338-341, 373-376, 399, 434-437, 465-468, 564, 702-706, 757, 761, 790-793, 803, 823-827, 834-837, 1093, 1097, 1129, 1242-1245, 1439-1447, 1588, 1672-1681, 1761-1764, 1775, 1780, 1785-1786, 1788, 1799, 1804, 1827, 1922-1941
R/tt_export.R                  472      44  90.68%   48, 65, 105-109, 142, 181-185, 249-262, 365-368, 383, 711, 720, 745-749, 916-919, 922, 953, 959
R/tt_from_df.R                   9       0  100.00%
R/tt_paginate.R                436      37  91.51%   48, 73, 110-118, 393, 515-518, 538-542, 692-695, 745-752, 821, 824, 834, 841, 844
R/tt_pos_and_access.R          571      43  92.47%   77, 79-81, 106, 166, 212-216, 262, 514, 516, 524, 530, 544, 554-557, 739, 750-754, 759-762, 789, 842-843, 855, 1021-1022, 1080-1108, 1388, 1465
R/tt_showmethods.R             144      21  85.42%   60, 97-120, 183, 209, 218, 226, 229-233, 326-327
R/tt_sort.R                     88       5  94.32%   223-226, 234
R/tt_toString.R                394      25  93.65%   118, 309, 357, 372, 382, 389, 392, 398-408, 497, 562, 568, 803-829
R/utils.R                       23       0  100.00%
R/validate_table_struct.R       84      12  85.71%   79-83, 92-93, 140, 150-151, 204-205
R/Viewer.R                      61       9  85.25%   51, 55, 65-69, 89, 123
TOTAL                         7802     698  91.05%

Diff against main

Filename                     Stmts    Miss  Cover
-------------------------  -------  ------  --------
R/00tabletrees.R              +113     +21  -1.84%
R/as_html.R                    +50      -1  +3.61%
R/colby_constructors.R         +73      +3  -0.08%
R/compare_rtables.R             +5      +6  -6.38%
R/index_footnotes.R            +12       0  +100.00%
R/make_split_fun.R            +119     +23  +80.67%
R/make_subset_expr.R           +30      +2  +1.03%
R/split_funs.R                 +84     +11  -0.01%
R/summary.R                    +32      +8  -2.42%
R/tree_accessors.R             +40     +11  -0.91%
R/tt_afun_utils.R              +66      +7  -0.54%
R/tt_compare_tables.R           +5       0  +0.44%
R/tt_compatibility.R           +97      +6  +1.13%
R/tt_dotabulation.R           +379     +52  -2.63%
R/tt_export.R                 +243     -29  +22.56%
R/tt_paginate.R                +53     +22  -4.57%
R/tt_pos_and_access.R          +51      +6  -0.42%
R/tt_showmethods.R             +23       0  +2.77%
R/tt_sort.R                     +7      -1  +1.73%
R/tt_toString.R                +89      +3  +0.87%
R/utils.R                      +12      -1  +9.09%
R/validate_table_struct.R      +84     +12  +85.71%
R/Viewer.R                      +5       0  +1.32%
TOTAL                        +1672    +161  -0.19%

Results for commit: 8ca8ef5

Minimum allowed coverage is 80%

♻️ This comment has been updated with latest results

[github-actions]

Unit Tests Summary

       1 files       24 suites   2m 32s ⏱️
   186 tests    186 ✔️ 0 💤 0 ❌
1 410 runs  1 410 ✔️ 0 💤 0 ❌

Results for commit 8ca8ef5.

♻️ This comment has been updated with latest results.

[shajoezhu]

test snap changing at insightsengineering/scda.test#67

[anajens]

I think this is too big of a breaking change right before the release. All of tern and TLG-C examples using ref_group will be affected.

Ideally the ref_group column should continue to be placed first by default. We just need to offer support to users who do not wish to use the default.

[Melkiades]

I think this is too big of a breaking change right before the release. All of tern and TLG-C examples using ref_group will be affected.

Ideally the ref_group column should continue to be placed first by default. We just need to offer support to users who do not wish to use the default.

I fully agree, I think it is better to give the possibility in tern to put the baseline somewhere else instead of the default

[clarkliming]

after simple evaluation, the ref_group is not affecting many templates. 12 templates in tlg-catalogue contains the ref_group argument, and only one snapshot is affected (ratet01, using ARM B as the reference group). Others all using the first level of arm as reference (so not changing snapshots).

Honestly I think this change should happen much earlier. With this update more flexibility is introduced and we finally have a consistent way of working with columns. Now factor levels is the only thing that changes the column order.

[Melkiades]

after simple evaluation, the ref_group is not affecting many templates. 12 templates in tlg-catalogue contains the ref_group argument, and only one snapshot is affected (ratet01, using ARM B as the reference group). Others all using the first level of arm as reference (so not changing snapshots).

Honestly I think this change should happen much earlier. With this update more flexibility is introduced and we finally have a consistent way of working with columns. Now factor levels is the only thing that changes the column order.

I understand your point of view, but I think it is better to keep the standard like it is and then have a split_fun in tern that takes care of any possible change. Here, we added 3 ways to do this https://github.com/insightsengineering/tern/pull/1087/files. @anajens mentioned the possibility of having a lot of columns, in that case having the ref group as standard would be very useful, but I do not know if this is in line with the standards (also last is the same practically). Maybe @barnett11 can help us with this issue!

[barnett11]

Thanks @Melkiades, I would expect Control group to be either first or last, but fixing it one place could be difficult. The standards do not mandate that control arm must come first, so i think we should allow flexibility for this change in order as we do for all of our templates. It is common, and STREAM has allowed this flexibility to users for some time

[Melkiades]

Just added blocked because I want to add the split functions at the same time so not to have changes in the TLG-c

[Melkiades]

btw changing ref_group positioning is affecting all of these, it was not apparent because the ref_group is by chance the first group: