| Type | Bug | Status | problem | Date | 16-Jun-2009 14:56 |
|---|---|---|---|---|---|
| Version | alpha 56 | Category | Unspecified | Submitted by | meijeru |
| Platform | All | Severity | minor | Priority | high |
| Summary | Spec of REMOVE-EACH says it returns remove count -- not so |
|---|---|
| Description |
In fact, remove-each returns the modified series at the argument position. This is the preferred behavior, and R2 compatible. The doc string needs to be changed to: {Removes values for each block that returns true. (Modifies)} |
| Example code |
>> source remove-each
remove-each: make native! [[
{Removes values for each block that returns true. Returns remove count. (Modifies)}
|
| Assigned to | n/a | Fixed in | - | Last Update | 18-Aug-2015 08:17 |
|---|
| Comments | |
|---|---|
|
(0001016)
BrianH 22-Jun-2009 16:35 |
In alpha 58 the behavior was changed to match the doc string, rather than changing the doc string to match the R2-compatible, preferred behavior.
Bug description changed to be more specific, and upgraded the bug from text to minor. |
|
(0003531)
rgchris 27-Feb-2013 22:46 |
The change appears to violate the rule of least surprise, though that could just be conditioning from having used it for many years. Is the new behaviour—returning the number of items removed—addressing a common problem? It'd also be straightforward to replicate and would seem (to me, at least) to be more deliberate:
length? block - length? remove-each item block [condition] On the other hand, we now have to assign a word to a block for all remove-each operations for which the block was to be returned. |
|
(0003533)
abolka 27-Feb-2013 23:46 |
I'd also prefer reverting this change back to R2's behaviour. I find the ability to chain remove-each calls far more important in practice than obtaining the number of removed elements. |
|
(0004659)
abolka 18-Aug-2015 08:17 |
Having had quite a few years with the changed behaviour, the situations where the removal count is helpful do exist (such as checking if something was removed with `if positive? remove-each ...`), but are _vastly_ outnumbered by use cases where R2's behaviour of returning the modified series is more useful.
This sentiment is shared by quite a few of R3's users, who describe their pain with the bad "chainability" of current REMOVE-EACH. We should therefore revert the default behaviour of REMOVE-EACH back to returning the modified argument series. If the removal count functionality is so desirable as to keep, I suggest adding a REMOVE-EACH/count refinement, which toggles REMOVE-EACH to returning the removal count. |
| Date | User | Field | Action | Change |
|---|---|---|---|---|
| 18-Aug-2015 08:17 | abolka | Comment : 0004659 | Added | - |
| 18-Aug-2015 08:13 | abolka | Category | Modified | => Unspecified |
| 18-Aug-2015 08:13 | abolka | Summary | Modified | spec of remove-each says it returns remove count -- not so => Spec of REMOVE-EACH says it returns remove count -- not so |
| 27-Feb-2013 23:46 | abolka | Comment : 0003533 | Added | - |
| 27-Feb-2013 22:46 | rgchris | Comment : 0003531 | Added | - |
| 22-Jun-2009 17:22 | BrianH | Comment : 0001016 | Modified | - |
| 22-Jun-2009 16:48 | BrianH | Comment : 0001016 | Modified | - |
| 22-Jun-2009 16:45 | BrianH | Description | Modified | - |
| 22-Jun-2009 16:45 | BrianH | Severity | Modified | text => minor |
| 22-Jun-2009 16:45 | BrianH | Description | Modified | - |
| 22-Jun-2009 16:41 | BrianH | Fixedin | Modified | alpha 58 => none |
| 22-Jun-2009 16:35 | BrianH | Comment : 0001016 | Added | - |
| 22-Jun-2009 16:30 | BrianH | Priority | Modified | normal => high |
| 22-Jun-2009 16:30 | BrianH | Status | Modified | built => problem |
| 20-Jun-2009 07:18 | carl | Status | Modified | reviewed => built |
| 20-Jun-2009 07:18 | carl | Fixedin | Modified | => alpha 58 |
| 17-Jun-2009 01:43 | BrianH | Description | Modified | - |
| 17-Jun-2009 01:42 | BrianH | Status | Modified | submitted => reviewed |
| 17-Jun-2009 01:42 | BrianH | Code | Modified | - |
| 16-Jun-2009 14:56 | meijeru | Ticket | Added | - |