Re: clojure.spec merge+or bug?

[Alex Whitt]

I got bitten by this today as well... makes it hard to add a predicate that 
destructures the map and compares its values. 

On Friday, August 17, 2018 at 4:35:12 PM UTC-4, Alex Miller wrote:
>
> With s/and, conformed values flow through the predicates. This allows you 
> to take advantage of structured values from early preds in later preds, so 
> you're not having to re-understand the structure. (There are cases where 
> having a non-flowing s/and would be useful and we've talked about adding 
> that since way back, still TBD). For single values, there's not much 
> difference, but it matters when you've got more structure, either from 
> regex (really s/& then) or s/or, etc.
>
> With s/merge the idea is to combine 2 or more map specs and "merge" their 
> specs, both during conform and gen. The mental picture I have is that the 
> same data flows into all the merged preds in parallel (whereas s/and snakes 
> them through in serial order). I don't have a good example at hand that 
> illustrates where is critical, but certainly in cases where you combining 
> s/keys with a collection view of map tuples with s/coll (sometimes called 
> "hybrid maps"), this is essential. 
>
>
>
> On Friday, August 17, 2018 at 2:36:23 PM UTC-5, Jenny Finkel wrote:
>>
>> Sorry, I meant to file a bug and even try to write a fix, but I was very 
>> pregnant at the time and then I gave birth so it sort of fell by the 
>> wayside. I'm interested in understanding why non-flowing behavior on 
>> conform is expected/desired behavior. I've found that when clojure design 
>> decisions go contrary to my intuitions, I usually learn a lot from 
>> understanding the design motivation. Would you mind explaining?
>> Cheers,
>>  Jenny
>> PS - I did try to google for an answer, and I found the thread where the 
>> docstring for merge got updated to reflect this, but I couldn't find an 
>> explanation.
>>
>> On Thu, Aug 16, 2018 at 7:00 PM, > 
>> wrote:
>>
>>> Thank you for explaining!
>>>
>>> Just in case, I opened a bug here 
>>> https://dev.clojure.org/jira/browse/CLJ-2388 
>>>
>>> On Thursday, August 16, 2018 at 6:47:30 PM UTC-7, Alex Miller wrote:

 The non-flowing behavior on conform is expected behavior.

 Failure to roundtrip conform then unform is a bug (so I'd so the bug 
 here is in unform).

 On a quick search, I don't believe this was filed, but I could have 
 missed it.

 On Thursday, August 16, 2018 at 8:28:54 PM UTC-5, shlomi...@gmail.com 
 wrote:
>
> Achhh, just spent the last few hours fighting this unexpected behavior 
> with s/merge, until I finally came to realize that this is what it was..
>
> I see this thread is quite old, did anyone open a bug for it as 
> mentioned above? 
>
> @Alex, you said this was the expected behavior, but then asked to open 
> a bug because it does not round-trip.. I am slightly confused, does the 
> problem lie in this "expected" behavior, or does it lie in s/unform? How 
> would such a bug be closed?
>
> Thanks,
> Shlomi
>
> On Thursday, June 1, 2017 at 6:24:25 AM UTC-7, Alex Miller wrote:
>>
>> You can file a bug on the s/merge unform - anything that doesn't 
>> roundtrip should be a bug.
>>
>> On the coll-of one, I thought that was just fixed in the latest 
>> spec.alpha release (see https://dev.clojure.org/jira/browse/CLJ-2076) 
>> - are you using latest there?
>>
> -- 
>>> You received this message because you are subscribed to the Google
>>> Groups "Clojure" group.
>>> To post to this group, send email to clo...@googlegroups.com 
>>> 
>>> Note that posts from new members are moderated - please be patient with 
>>> your first post.
>>> To unsubscribe from this group, send email to
>>> clo...@googlegroups.com 
>>> For more options, visit this group at
>>> http://groups.google.com/group/clojure?hl=en
>>> --- 
>>> You received this message because you are subscribed to a topic in the 
>>> Google Groups "Clojure" group.
>>> To unsubscribe from this topic, visit 
>>> https://groups.google.com/d/topic/clojure/r8WO24rHsi0/unsubscribe.
>>> To unsubscribe from this group and all its topics, send an email to 
>>> clo...@googlegroups.com .
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
To view this discussion on 

Re: clojure.spec merge+or bug?

[Alex Miller]

With s/and, conformed values flow through the predicates. This allows you 
to take advantage of structured values from early preds in later preds, so 
you're not having to re-understand the structure. (There are cases where 
having a non-flowing s/and would be useful and we've talked about adding 
that since way back, still TBD). For single values, there's not much 
difference, but it matters when you've got more structure, either from 
regex (really s/& then) or s/or, etc.

With s/merge the idea is to combine 2 or more map specs and "merge" their 
specs, both during conform and gen. The mental picture I have is that the 
same data flows into all the merged preds in parallel (whereas s/and snakes 
them through in serial order). I don't have a good example at hand that 
illustrates where is critical, but certainly in cases where you combining 
s/keys with a collection view of map tuples with s/coll (sometimes called 
"hybrid maps"), this is essential. 



On Friday, August 17, 2018 at 2:36:23 PM UTC-5, Jenny Finkel wrote:
>
> Sorry, I meant to file a bug and even try to write a fix, but I was very 
> pregnant at the time and then I gave birth so it sort of fell by the 
> wayside. I'm interested in understanding why non-flowing behavior on 
> conform is expected/desired behavior. I've found that when clojure design 
> decisions go contrary to my intuitions, I usually learn a lot from 
> understanding the design motivation. Would you mind explaining?
> Cheers,
>  Jenny
> PS - I did try to google for an answer, and I found the thread where the 
> docstring for merge got updated to reflect this, but I couldn't find an 
> explanation.
>
> On Thu, Aug 16, 2018 at 7:00 PM,  wrote:
>
>> Thank you for explaining!
>>
>> Just in case, I opened a bug here 
>> https://dev.clojure.org/jira/browse/CLJ-2388 
>>
>> On Thursday, August 16, 2018 at 6:47:30 PM UTC-7, Alex Miller wrote:
>>>
>>> The non-flowing behavior on conform is expected behavior.
>>>
>>> Failure to roundtrip conform then unform is a bug (so I'd so the bug 
>>> here is in unform).
>>>
>>> On a quick search, I don't believe this was filed, but I could have 
>>> missed it.
>>>
>>> On Thursday, August 16, 2018 at 8:28:54 PM UTC-5, shlomi...@gmail.com 
>>> wrote:

 Achhh, just spent the last few hours fighting this unexpected behavior 
 with s/merge, until I finally came to realize that this is what it was..

 I see this thread is quite old, did anyone open a bug for it as 
 mentioned above? 

 @Alex, you said this was the expected behavior, but then asked to open 
 a bug because it does not round-trip.. I am slightly confused, does the 
 problem lie in this "expected" behavior, or does it lie in s/unform? How 
 would such a bug be closed?

 Thanks,
 Shlomi

 On Thursday, June 1, 2017 at 6:24:25 AM UTC-7, Alex Miller wrote:
>
> You can file a bug on the s/merge unform - anything that doesn't 
> roundtrip should be a bug.
>
> On the coll-of one, I thought that was just fixed in the latest 
> spec.alpha release (see https://dev.clojure.org/jira/browse/CLJ-2076) 
> - are you using latest there?
>
 -- 
>> You received this message because you are subscribed to the Google
>> Groups "Clojure" group.
>> To post to this group, send email to clojure@googlegroups.com
>> Note that posts from new members are moderated - please be patient with 
>> your first post.
>> To unsubscribe from this group, send email to
>> clojure+unsubscr...@googlegroups.com
>> For more options, visit this group at
>> http://groups.google.com/group/clojure?hl=en
>> --- 
>> You received this message because you are subscribed to a topic in the 
>> Google Groups "Clojure" group.
>> To unsubscribe from this topic, visit 
>> https://groups.google.com/d/topic/clojure/r8WO24rHsi0/unsubscribe.
>> To unsubscribe from this group and all its topics, send an email to 
>> clojure+unsubscr...@googlegroups.com.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[Jenny Finkel]

Sorry, I meant to file a bug and even try to write a fix, but I was very
pregnant at the time and then I gave birth so it sort of fell by the
wayside. I'm interested in understanding why non-flowing behavior on
conform is expected/desired behavior. I've found that when clojure design
decisions go contrary to my intuitions, I usually learn a lot from
understanding the design motivation. Would you mind explaining?
Cheers,
 Jenny
PS - I did try to google for an answer, and I found the thread where the
docstring for merge got updated to reflect this, but I couldn't find an
explanation.

On Thu, Aug 16, 2018 at 7:00 PM,  wrote:

> Thank you for explaining!
>
> Just in case, I opened a bug here https://dev.clojure.org/jira/
> browse/CLJ-2388
>
> On Thursday, August 16, 2018 at 6:47:30 PM UTC-7, Alex Miller wrote:
>>
>> The non-flowing behavior on conform is expected behavior.
>>
>> Failure to roundtrip conform then unform is a bug (so I'd so the bug here
>> is in unform).
>>
>> On a quick search, I don't believe this was filed, but I could have
>> missed it.
>>
>> On Thursday, August 16, 2018 at 8:28:54 PM UTC-5, shlomi...@gmail.com
>> wrote:
>>>
>>> Achhh, just spent the last few hours fighting this unexpected behavior
>>> with s/merge, until I finally came to realize that this is what it was..
>>>
>>> I see this thread is quite old, did anyone open a bug for it as
>>> mentioned above?
>>>
>>> @Alex, you said this was the expected behavior, but then asked to open a
>>> bug because it does not round-trip.. I am slightly confused, does the
>>> problem lie in this "expected" behavior, or does it lie in s/unform? How
>>> would such a bug be closed?
>>>
>>> Thanks,
>>> Shlomi
>>>
>>> On Thursday, June 1, 2017 at 6:24:25 AM UTC-7, Alex Miller wrote:

 You can file a bug on the s/merge unform - anything that doesn't
 roundtrip should be a bug.

 On the coll-of one, I thought that was just fixed in the latest
 spec.alpha release (see https://dev.clojure.org/jira/browse/CLJ-2076)
 - are you using latest there?

>>> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to clojure@googlegroups.com
> Note that posts from new members are moderated - please be patient with
> your first post.
> To unsubscribe from this group, send email to
> clojure+unsubscr...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en
> ---
> You received this message because you are subscribed to a topic in the
> Google Groups "Clojure" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/
> topic/clojure/r8WO24rHsi0/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> clojure+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[shlomivaknin]

Thank you for explaining!

Just in case, I opened a bug here 
https://dev.clojure.org/jira/browse/CLJ-2388 

On Thursday, August 16, 2018 at 6:47:30 PM UTC-7, Alex Miller wrote:
>
> The non-flowing behavior on conform is expected behavior.
>
> Failure to roundtrip conform then unform is a bug (so I'd so the bug here 
> is in unform).
>
> On a quick search, I don't believe this was filed, but I could have missed 
> it.
>
> On Thursday, August 16, 2018 at 8:28:54 PM UTC-5, shlomi...@gmail.com 
>  wrote:
>>
>> Achhh, just spent the last few hours fighting this unexpected behavior 
>> with s/merge, until I finally came to realize that this is what it was..
>>
>> I see this thread is quite old, did anyone open a bug for it as mentioned 
>> above? 
>>
>> @Alex, you said this was the expected behavior, but then asked to open a 
>> bug because it does not round-trip.. I am slightly confused, does the 
>> problem lie in this "expected" behavior, or does it lie in s/unform? How 
>> would such a bug be closed?
>>
>> Thanks,
>> Shlomi
>>
>> On Thursday, June 1, 2017 at 6:24:25 AM UTC-7, Alex Miller wrote:
>>>
>>> You can file a bug on the s/merge unform - anything that doesn't 
>>> roundtrip should be a bug.
>>>
>>> On the coll-of one, I thought that was just fixed in the latest 
>>> spec.alpha release (see https://dev.clojure.org/jira/browse/CLJ-2076) - 
>>> are you using latest there?
>>>
>>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[Alex Miller]

The non-flowing behavior on conform is expected behavior.

Failure to roundtrip conform then unform is a bug (so I'd so the bug here 
is in unform).

On a quick search, I don't believe this was filed, but I could have missed 
it.

On Thursday, August 16, 2018 at 8:28:54 PM UTC-5, shlomivak...@gmail.com 
wrote:
>
> Achhh, just spent the last few hours fighting this unexpected behavior 
> with s/merge, until I finally came to realize that this is what it was..
>
> I see this thread is quite old, did anyone open a bug for it as mentioned 
> above? 
>
> @Alex, you said this was the expected behavior, but then asked to open a 
> bug because it does not round-trip.. I am slightly confused, does the 
> problem lie in this "expected" behavior, or does it lie in s/unform? How 
> would such a bug be closed?
>
> Thanks,
> Shlomi
>
> On Thursday, June 1, 2017 at 6:24:25 AM UTC-7, Alex Miller wrote:
>>
>> You can file a bug on the s/merge unform - anything that doesn't 
>> roundtrip should be a bug.
>>
>> On the coll-of one, I thought that was just fixed in the latest 
>> spec.alpha release (see https://dev.clojure.org/jira/browse/CLJ-2076) - 
>> are you using latest there?
>>
>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[shlomivaknin]

Achhh, just spent the last few hours fighting this unexpected behavior with 
s/merge, until I finally came to realize that this is what it was..

I see this thread is quite old, did anyone open a bug for it as mentioned 
above? 

@Alex, you said this was the expected behavior, but then asked to open a 
bug because it does not round-trip.. I am slightly confused, does the 
problem lie in this "expected" behavior, or does it lie in s/unform? How 
would such a bug be closed?

Thanks,
Shlomi

On Thursday, June 1, 2017 at 6:24:25 AM UTC-7, Alex Miller wrote:
>
> You can file a bug on the s/merge unform - anything that doesn't roundtrip 
> should be a bug.
>
> On the coll-of one, I thought that was just fixed in the latest spec.alpha 
> release (see https://dev.clojure.org/jira/browse/CLJ-2076) - are you 
> using latest there?
>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[Alex Miller]

You can file a bug on the s/merge unform - anything that doesn't roundtrip 
should be a bug.

On the coll-of one, I thought that was just fixed in the latest spec.alpha 
release (see https://dev.clojure.org/jira/browse/CLJ-2076) - are you using 
latest there?

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[Jenny Finkel]

It also seems to come up with coll-of + or:

user> (s/def ::a (s/or :even even? :odd odd?))
user> (s/def ::c (s/coll-of ::a))
user> (s/conform ::c [1 2 3 4])
[[:odd 1] [:even 2] [:odd 3] [:even 4]]
user> (s/unform ::c (s/conform ::c [1 2 3 4]))
[[:odd 1] [:even 2] [:odd 3] [:even 4]]

It looks like coll-of calls every-impl which just has identity as it's 
unform fn.
-Jenny

On Wednesday, May 31, 2017 at 9:37:43 PM UTC-7, Jenny Finkel wrote:
>
> I think I found a bug in spec when combining merge and or. Basically, when 
> you conform a map where one of the keys has an or, and the spec comes from 
> a clojure.spec/merge, one of the underlying keys will conform it, while the 
> others don't, and then when the results get merged together you can end up 
> with the unconformed version in the result:
>
> user> (require '[clojure.spec :as s])
> user> (s/def ::a (s/or :even even? :odd odd?))
> user> (s/def ::b (s/or :even even? :odd odd?))
> user> (s/def ::m1 (s/keys :req-un [::a]))
> user> (s/def ::m2 (s/keys :req-un [::b]))
> user> (s/def ::mm (s/merge ::m1 ::m2))
> user> (s/valid? ::mm {:a 1 :b 2})
> true
> user> (s/conform ::mm {:a 1 :b 2})
> {:a 1, :b [:even 2]}
> user> (s/unform ::mm {:a [:odd 1] :b [:even 2]})
> {:a 1, :b [:even 2]}
> user> (s/unform ::mm (s/conform ::mm {:a 1 :b 2}))
> UnsupportedOperationException nth not supported on this type: Long 
>  clojure.lang.RT.nthFrom (RT.java:962)
>
> I guess that valid? checks if it satisfies all the merged specs, and 
> conform conforms on each and merges, and then unform can end up with a 
> result that it can't unform (and similarly, if you give unform a properly 
> conformed thing it can't unform properly due to how it merges results as 
> well). I think the fix would be an update to clojure.spec/merge to make it 
> smarter about which keys to keep from each map on conforming and unforming, 
> though looking at the current code I don't think it's an easy fix.
>
> -Jenny
>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[Jenny Finkel]

thanks for the fast reply! do you think it will always be this way? it does
seem to violate the expected conform/unform relationship.

On Wed, May 31, 2017 at 9:51 PM, Alex Miller  wrote:

> This is actually the expected result. s/merge doesn't flow like s/and -
> only the conformed version of the last map spec in the merge is used. There
> are thus some unexpected results in combination with the -un options as
> they are the only link towards conforming.
>
> One thing you could try instead is to use s/and which does flow and might
> give you some of the behavior you're looking for (but won't gen as well).
>
> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to clojure@googlegroups.com
> Note that posts from new members are moderated - please be patient with
> your first post.
> To unsubscribe from this group, send email to
> clojure+unsubscr...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en
> ---
> You received this message because you are subscribed to a topic in the
> Google Groups "Clojure" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/
> topic/clojure/r8WO24rHsi0/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> clojure+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

clojure.spec merge+or bug?

[Alex Miller]

This is actually the expected result. s/merge doesn't flow like s/and - only 
the conformed version of the last map spec in the merge is used. There are thus 
some unexpected results in combination with the -un options as they are the 
only link towards conforming.

One thing you could try instead is to use s/and which does flow and might give 
you some of the behavior you're looking for (but won't gen as well).

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: clojure.spec merge+or bug?

[Jenny Finkel]

PS - I just realize I wasn't using the latest version of spec, as is 
evident from my require, but I just tried again with the latest and it 
doesn't change the result. And I should have mentioned that I'm happy to 
take a stab at a fix, assuming I'm correct that this is a bug.

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

clojure.spec merge+or bug?

[Jenny Finkel]

I think I found a bug in spec when combining merge and or. Basically, when 
you conform a map where one of the keys has an or, and the spec comes from 
a clojure.spec/merge, one of the underlying keys will conform it, while the 
others don't, and then when the results get merged together you can end up 
with the unconformed version in the result:

user> (require '[clojure.spec :as s])
user> (s/def ::a (s/or :even even? :odd odd?))
user> (s/def ::b (s/or :even even? :odd odd?))
user> (s/def ::m1 (s/keys :req-un [::a]))
user> (s/def ::m2 (s/keys :req-un [::b]))
user> (s/def ::mm (s/merge ::m1 ::m2))
user> (s/valid? ::mm {:a 1 :b 2})
true
user> (s/conform ::mm {:a 1 :b 2})
{:a 1, :b [:even 2]}
user> (s/unform ::mm {:a [:odd 1] :b [:even 2]})
{:a 1, :b [:even 2]}
user> (s/unform ::mm (s/conform ::mm {:a 1 :b 2}))
UnsupportedOperationException nth not supported on this type: Long 
 clojure.lang.RT.nthFrom (RT.java:962)

I guess that valid? checks if it satisfies all the merged specs, and 
conform conforms on each and merges, and then unform can end up with a 
result that it can't unform (and similarly, if you give unform a properly 
conformed thing it can't unform properly due to how it merges results as 
well). I think the fix would be an update to clojure.spec/merge to make it 
smarter about which keys to keep from each map on conforming and unforming, 
though looking at the current code I don't think it's an easy fix.

-Jenny

-- 
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your 
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.