« first day (2252 days earlier)      last day (632 days later) » 

6:58 AM
@ilian @halirutan @gwr while this topic is indeed not strictly by the SE rules, we know we bend them a lot here. I'd suggest @Nasser to undelete it and at least let it be discussed. E.g. I didn't have much time during weekend and missed that one. And I'm for keeping it alive, at least as part of 'backward incompatible changes' topic.
@ilian First of all, we could not be more happy to have a direct feedback from developers the more that quite a lot of it comes from their free time. Otoh I'm not surprised that users may find it confusing that some 'clear bug reports' are kept and some are (only apparently) 'attacked as off topic'. And while I understand both sides we should try to be more pro community (and we usually are)
Finally, couple of developers told me, if you want the bug get fixed, make a noise.
So we should try to rephrase this question to 'what to do in 11.3 to make them evalaute' or whatever that is strictly on topic just to make it stay. @halirutan @gwr @Nasser @ilian
But I'm open to discussion too.
2 hours later…
9:27 AM
@Kuba @halirutan @gwr @Nasser Maybe it would fit better with the Stack Exchange paradigm if we made a question like this: "Are there any known regressions in the new 11.3 release?" and then Nasser could answer that question with that excellent list. What do you think?
@C.E. I like this but one could still argue that it is as broad/off topic as the former one.
But it gives us a nice place to gather them which is what I like about that.
Not really what SE if for though
@Kuba I agree that it isn't perfect, but it is perhaps a step in the right direction. As you have already said, we do tend to bend the rules when it is beneficial for us. This suggestion is more about how to bend the rules in the best way.
Yep +1
9:59 AM
What do you get in 11.1/2/3 for:
URLDownload["https://httpbin.org/status/401", Interactive -> False]
For <11.2 I get an authentication dialog which should not happen.

For 11.2 Win7 I get File[_] to a file that does not exist despite 401

For 11.2 Win10 I get a bunch of Lookup/CURLLink errors and a Failure

For 11.3 I get a File[_] which does not exist -.-
11.2 Linux an empty File
10:19 AM
I don't get it. Do people use those new web functions? I do basic stuff and expect to handle exceptions nicely and what I end up doing is writing bug reports each time I take another from URL* family for testing.
10:40 AM
@ilian I apologize for having use such "drastic" idiom but I still do find the question whether WRI employees do get a vote for closing bug-related questions valid in principle. To be clear, I greatly appreciate your effort and having you around does make M.SE more valuable.
@Kuba @halirutan @ C.E. @Nasser I know the rules for SE and have seen the bug-related-tag-warnings. :) But coming from economics and business, I would say: If such a public place to track bugs and "regressions" were not according to SE's rules then there should be another place for it as having such a thing should clearly benefit all sides.
In seeing a bug tag and a Case-reference here, I will not report the issue on my own since I know it has been reported already. This will save time for me and WRI at the same time. Also this open discussion and voting to WRI is valuable as it shows user interest - without additional cost.
While bug-tags can be painful and may not always be related with "anything we can do immediately" (e.g. "need to contact WRI" in the close vote) it most of the time is of interest to the community here (at least with Nasser's post it was imo) and there are sometimes deeper issues: Something has obviously changed that prevented integration rules to work as usual - maybe this relates to other functionality. Now, we will never know...
Last sentence: Think about it this way, Maple and MatLab have never gotten an SE of their own while users still tried on Area 51 - WRI here is far ahead and should benefit even with a bug report too many here for show?
@gwr Nasser's case is an outlier. He is a long member of the community and this post shouldn't have been removed. Maybe changed to fit the rules but even without it is nice that someone took the time to inspect these integrals.
Usually when bug-tagged questions are closed it's a beginners mistake.
@halirutan I was surprised that WRI is not using the list -- which I believe is coming from Rubi -- as a kind of automatic integration test (in the software development way of usage)?
@gwr Yes, me too. I would be really curious to which extend Wolfram uses serious unit testing at all.
There could have been open communication by marketing then, e.g. we changed this and that for some reason and this may leed to some preliminary regressions but will improve Mathematica in the end... or something similiar.
@Nasser please do undelete your post and maybe give just a couple of examples with linking to your full report. Slight rewording and hinting at "workarounds" (e.g. use Rubi - which was originally written in Mathematica afaik) or else should make this a viable question imho.
10:57 AM
@gwr I really like changes, updates and fixes for Hash in 11.3 but we know how this ended up: mathematica.stackexchange.com/questions/167884/…
So yes, it is a opportunity for a good marketing but it is for a long time and it is not used. Except of prompt updates by Ilian, Itai and others to existing topics.
They are not company policy driven though, but a private effort which I'm grateful for.
@kuba Yes, I can clearly see that which is why I have taken back a bit of my "strike out" there. I am an old fashioned Popperian here: open and honest critique makes the world a better place - or at least the "Wolfram Universe" which probably will be the new title for M.SE with Mathematica being a "legacy product" :)
12:01 PM
@Nasser, I can undelete your post, but I would very much prefer that you undelete it yourself. I think this is a case where we can bend the rules a bit; maybe just pose it in question format so that it still somewhat fits SE?
12:26 PM
@J.M. others. I undeleted the question per request. Please feel free to make any other changes as you like to make the question better worded. Thanks
2 hours later…
2:53 PM
@Kuba I see myself as a big user of all these web functions. So... I thank you for being in advance testing them...
3 hours later…
5:35 PM
Anyone know how old System`Private`Arguments is? I'd be interested in using it if it's from 10.0 at least.
6:06 PM
@b3m2a1 - works in version 9 at least
I might expect it to work as far back as version 7, but I can't test that
@JasonB. thanks. That's a highly convenient function. I wish those developer-oriented functions were documented and in a less suspicious context...
I like that one because it will give me the message for the wrong number of arguments, but it doesn't always distinguish rules as Options, putting them in the arguments list
In[32]:= Options[func] = {"foo" -> bar}

Out[32]= {"foo" -> bar}

In[33]:= System`Private`Arguments[func[a, "foo" -> bar], {2, 3}]

Out[33]= {{a, "foo" -> bar}, {}}
@JasonB. probably needs an OptionsPattern[] registered to work properly... OptionsPattern (and OptionValue) is definitely a little piece of black-magic.
okay, i'll go home now
er, idk - the options pattern goes in func down values
Actually nope it's kinda broken either way...
func // Clear;
Options[func] = {"foo" -> bar};
func[a_, ops : OptionsPattern[]] := Null
System`Private`Arguments[func[a, "foo" -> bar], {2, 3}]

{{a, "foo" -> bar}, {}}
6:17 PM
but Arguments seems like it doesn't check down values
@b3m2a1 version 8 has it. version 5 definitely doesn't. ;)
What OS do you have you can run version 5 on?
@J.M. I've never used before 9 (and all my code is littered with Association) but that's good to know.
@JasonB. actually, whenever I have to use version 5, I ask a friend if he can let me use his Windows computer remotely.
I wonder if Arguments was part of the OptionsPattern push in v6
6:19 PM
version 5 definitely won't install in current Linux distros, due to mesa and other junk I didn't want to investigate
@JasonB. another one I like but always forget about:
System`Private`ArgumentsWithRules[func[a, c, b, "nope" -> "nope", "foo" -> bar], {2, 3}]
Provides first-order options checking too which is convenient.
ah, that's the one I remember using for stuff with options.
At one point or another I want to make my packages look like WRI's stuff with all that nice syntax highlighting and options checking but I've just never gotten around to it... I even wrote a nicer helper function for it but I'm still working out how I want to include my list of annotations.
2 hours later…
8:18 PM
@gwr @C.E. @halirutan @J.M. @Nasser @Szabolcs and everyone interested. I moved the discussion from comments about Integrate in 11.3 to a separate channel:

8:35 PM
@Kuba Good call, it was getting messy already. (Considering how tolerant we are of comment discussions, more than most other SE sites.)
@J.M. I left two comments in OP, not 100% sure it is fair but should give a quick overview. So feel free to delete / tell me to delete.
9:07 PM
@Kuba any idea what this font syntax means:

".SF NS Text"
Just some OSX specific thing
@Kuba Thanks, that was necessary.
23 hours ago, by gwr
Personally I find being informed about such kind of "changes" rather interesting and good to know. After all, this community is rather fair about assigning "bug" tags and informing about their resolution. Not making such things transparent to me looks like too much self-censorship?
@gwr @Nasser @halirutan @ilian and others:
I agree that this is something interesting to discuss. But it a topic for a discussion forum (like Wolfram Community), and really not a good place for a Q/A site.
There's also the very real problem that people will periodically attempt to use this site as a bug reporting tool, which is completely misguided. It is in everyone's interest to report problem to WRI to give them a chance to be fixed. What often happens instead is that people will say "it's not my job to write reports" (despite having written one here) or that "they won't fix it anyway, why report it", etc.
When I saw Nasser's post, I assumed that he has already reported the problems, but then there's also the issue of setting an example. If high rep users do this, everyone else will do this. If bug-discussions (not questions) are accepted, why aren't other discussions accepted?
@gwr So to sum up, I never saw this issue as censorship, as trying to let WRI save face, or anything even remotely similar. It's as simple as a bug report having no place on a QA site. Remember that M.SE is run by StackExchange Inc, and it's their decision to use and enforce this format.
Is there a need to also have a free discussion forum? I think yes, but this is not the place for it.
9:24 PM
@Szabolcs nominally this is what Wolfram Community would be for but I've generally found the response quality there to be lower than on SE and the site to be super buggy so I tend to only go there when I want to push for community support of some feature

« first day (2252 days earlier)      last day (632 days later) »