# [RFC Proposal] var keyword deprecation/removal - Thread: https://externals.io/message/91330 - Original archive: https://news-web.php.net/php.internals/91330 - Started by Fleshgrinder on 2016-02-19T20:05:09+00:00 - Messages: 9 - Reply by email to internals@lists.php.net - More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt --- ## #91330 — Fleshgrinder — 2016-02-19T20:05:09+00:00 URL: https://externals.io/message/91330#91330 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 > This is all generic advice which is nice and well if we would be > designing language anew. As it is, we are not - we already have > lots of code using var. For code not using var, removing var does > not do anything. For code using var, removing var means no upgrade. > It is unnecessary breakage for the sake of feeling that we follow > the advice somebody put on the website. That feeling is not worth > very real work that people would have to do to achieve what they > already have now. I would not say that the information that the Nielsen Norman Group puts on their website falls in the category of "somebody putting something on a website". https://www.nngroup.com/about/ https://en.wikipedia.org/wiki/Nielsen_Norman_Group Additionally there is other scientific work that supports what I stated, e.g.: http://www.eecs.yorku.ca/research/techreports/1999/CS-1999-08.pdf > Yes, PHP has aliases. Nothing wrong with that, provided you bother > to look in the manual once, and then you are enlightened forever. > [...] It only wastes their time if they don't read the manual but > instead chat on Stackoverflow :) Which I have nothing against, it's > just not the reason to introduce BC breaks because somebody asked a > question about it. A lot of things are wrong with this (see above reasons) and the RTFM argument is a thought-terminating cliché. Duplication harms the overall design of a language, it results in a maintainability burden for the developers, hell, even this discussion is only happening and wasting our time due to these duplications. I am not saying that your arguments about breaking changes and more complicated upgrades for a small user base are wrong; I am saying that these are weak arguments compared to the scientific facts and the overall progress for the PHP programming language. After all, we are talking about an upgrade path of several years. Anyone should be able to adopt by that time and they probably have other problems until then with their old systems. - -- Richard "Fleshgrinder" Fussenegger -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBCAAGBQJWx3V1AAoJEOKkKcqFPVVrc4cQAINj1zV9lXZAaMGCHrTWjtK8 RC04cjsYzG0XMUa+8wexxmQCkgMSBkRLOWaDa8VyCTbUabHSHi70Zf2mj0qyFQcM rv6DxYVorWDx5ty0odnv+8vgnBra2LHc4OntRGe0GWsM0Z4aQ2QfriMzaAt+diBn gSCC5yPxfE3jjqi4TYuT7eshQ3Kz0SeTAarY19fApi7f80JDwChHeerUiGFMZE3x GUUJi02NeKUnfqliAM0k9VGgAP4qGjmPiVFWjoQKZ9rYPvGRqztLO7c2V+65DlY7 He/vibZDsmvbQZXVcszi2cWNhhzn43XXTwF+jiFuKhlXNLwJJDcJbanfJtQJpEiI MB5adPIROxTtgJdrOBTZ8PMT0IKk4P1i19IUqL6sne4AJtcn3UYEoNUiUiO2TJfP 1zgNfTOAnlmf14jtij2g1wj8XPscjA38zo/2QWzGxadqgsQX6m2NgqlzMV+D6WbH C5KToC5EiF/WNWd/vsViYLCEPusuFzHDx8qWFH0ZiNBdhsMTX6yB+Rt5mKZ2hauY o4+5z4308JPrI8Ao6oroVJ3TqiGN7R8nWUlE3J5/v9psgYVdFMyIe1vXmhtROBo9 N382klHxZHVISis0290HnmdlP8t/g6AevdF3uGHbLEUA8Jzs9LwzrAlUtSBkCNlx 3Yeby46U03XVb3zCZzm3 =1iGy -----END PGP SIGNATURE --- ## #91331 — Stanislav Malyshev — 2016-02-19T20:49:12+00:00 URL: https://externals.io/message/91330#91331 · In reply to: #91330 Hi! > I would not say that the information that the Nielsen Norman Group > puts on their website falls in the category of "somebody putting > something on a website". I think you are trying for an argument of "I agree with some guys that I consider being an authority so you should agree with me". It works only if these guys are established as unerring God-like authority _and_ they spoke on the particular case we are discussing. I don't think either of this is true. So let's dispense with authority appeals for now. > A lot of things are wrong with this (see above reasons) and the RTFM > argument is a thought-terminating cliché. Duplication harms the It is cliche because it's true. You do not need any deep thought to know what aliases are, there is no deep insight to be had there, there is just a thing to know - these two things are two ways of saying the same. It's not really hard or "confusing". > overall design of a language, it results in a maintainability burden > for the developers, hell, even this discussion is only happening and > wasting our time due to these duplications. Wait, so the fact of you arguing for removal of aliases is the proof of you being right? That's neat. -- Stas Malyshev smalyshev@gmail.com --- ## #91332 — Fleshgrinder — 2016-02-19T21:09:42+00:00 URL: https://externals.io/message/91330#91332 · In reply to: #91331 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 > I think you are trying for an argument of "I agree with some guys > that I consider being an authority so you should agree with me". It > works only if these guys are established as unerring God-like > authority _and_ they spoke on the particular case we are > discussing. I don't think either of this is true. So let's dispense > with authority appeals for now. No, I just wanted to remark that it is not just an opinion based blog post that I was referring to and that the link points to a _valid_ argument. > It is cliche because it's true. You do not need any deep thought to > know what aliases are, there is no deep insight to be had there, > there is just a thing to know - these two things are two ways of > saying the same. It's not really hard or "confusing". Old cellphones were shipped with a user manual that contained precise instructions on how to deal with the installed OS. New smartphones do not contain a user manual because the OS is so intuitive that nobody has a need for them. I never read the manual for any of those old cellphones and was still able to achieve my goals. However, other people apparently had problems and that is why a manual was shipped along with them. Do not apply your intelligence and knowledge on others and approach topics by weighting pros and cons and not opinions. > Wait, so the fact of you arguing for removal of aliases is the > proof of you being right? That's neat. No and I did not imply this, again, we are collecting arguments and I am on the pro side so I contribute pro arguments. Your comments on the other hand are passive aggressive and do not contribute to the overall discussion because you simply say no and do not even allow a discussion. I received your actual arguments and I acknowledge them but _I personally_ consider them weak (if that is true cannot be judged by either of us) and _I personally_ think that the arguments others and me brought to the table are more extensive and thus add more to the pro side even if your arguments count fully on the con side. - -- Richard "Fleshgrinder" Fussenegger -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQIcBAEBCAAGBQJWx4SWAAoJEOKkKcqFPVVraPcQAJ7h/R4qmB/BhIgsVIWUVdYY TrMffjrO1Mz8DaFhVsjjDhjj2oQ9OHvcGOTZ/WRVsMRxO6wh1C3QCuQFrE/uCO4z 2oLV54MRcs4DyNUzIflS/iPrpfLI1aplQZDlDhB1Q470rr7sj3p69dqAhnHEIQdE jD4DePDIRIe4b7tfOYfC4HF3mGXyufisiLPGXrYxAEKNmy+4DewnhnN7nnP2Zst4 gXxzDtzNlQbQS/R3TFu8xPZZ4S8e71wn0KdNhSULgM5m4rDQBGFisgHuJe8eLTxJ 9Y0XLNZh4L7uUcAyfi6yqZ61krhvZcm2J234xxgS6425NKyQ+qAI7kWshv2dNZdi Xwm2asgOOmwmNpXhO2JOJpf+1K459nL8k7Po2qckdFqD332iEkEKirNZp/YpDuzx gClz4oriS3ygH9AVgvy+EtZ+dUkzdLxXFWtN6MX1UzQ1zYRc1gIIgiTvLuoraupJ d2Qp4ZbiFEtS3YQm+lInpDgarIKGqNQ1zXMEnRogENy8UC0FtzYP3n0Sx+J08Sxd CeQ7/9aLneONFOTKYK/Ea+3YyXVjBWtq7JJqVZdwearkqaUn5+TTNSczAyfV968I GB96lk2R4voTfrD9nTmrPot5FPZA/acGKPFpvSQNFjNi/AG/UNR+WTsK4JnWpq2I gUE4PXwx3AFxZBHCh5Zg =HMPP -----END PGP SIGNATURE --- ## #91345 — Stanislav Malyshev — 2016-02-22T20:12:23+00:00 URL: https://externals.io/message/91330#91345 · In reply to: #91332 Hi! > Old cellphones were shipped with a user manual that contained precise > instructions on how to deal with the installed OS. You don't really need a whole manual to know two things are the same. You only need one line from that manual. It's a minimal effort, well within expected of what may be required of a person learning new language - I think reading a couple of lines is not excessive. > New smartphones do not contain a user manual because the OS is so > intuitive that nobody has a need for them. Or so the marketing team would love us to believe ;) If you think a programming language somehow can be so "intuitive" that you never need to actually know anything to use it - you're in for a somewhat unpleasant surprise, unfortunately. We can make learning it easi-er - and having aliases is part of it - but we can never make it so easy as never having to actually touch the manual or any other learning media. > discussion because you simply say no and do not even allow a That is not correct. I say no with substantiation - namely, that removing aliases would cause code breakage and would not add anything to actual functionality. Your argument seems to be generic "redundancy is bad" argument, and treating somebody asking questions on stackoverflow as evidence that we have a problem. Both are wrong - redudancy can be both good and bad, and in our case I think it is good because it lets people continue to rely on their previous experience both in PHP and other languages. Also, somebody asking questions is not a reason to change the language - there always will be people asking questions, and that's fine as long as we have good easily accessible answers, which we do. -- Stas Malyshev smalyshev@gmail.com --- ## #91379 — Fleshgrinder — 2016-02-24T17:41:29+00:00 URL: https://externals.io/message/91330#91379 · In reply to: #91345 > Hi! Hey there. :) > > […] > > You don't really need a whole manual to know two things are the same. > You only need one line from that manual. It's a minimal effort, well > within expected of what may be required of a person learning new > language - I think reading a couple of lines is not excessive. > > > […] > > Or so the marketing team would love us to believe ;) > If you think a programming language somehow can be so "intuitive" that > you never need to actually know anything to use it - you're in for a > somewhat unpleasant surprise, unfortunately. We can make learning it > easi-er - and having aliases is part of it - but we can never make it so > easy as never having to actually touch the manual or any other learning > media. My example seems to be bad or to abstract. I am not thinking that a programming language can become that intuitive; the contrary. Programming languages are already so complicated that it is harmful to start littering them with thousands of ways to achieve one thing. It also makes users uncertain when the manual entry says that something is an alias. They fear that it might be removed. They create project rules that forbid the usage of aliases ... Science shows that it is harmful, let's clean it up! > > […] > > That is not correct. I say no with substantiation - namely, that > removing aliases would cause code breakage and would not add anything to > actual functionality. Your argument seems to be generic "redundancy is > bad" argument, and treating somebody asking questions on stackoverflow > as evidence that we have a problem. Both are wrong - redudancy can be > both good and bad, and in our case I think it is good because it lets > people continue to rely on their previous experience both in PHP and > other languages. Also, somebody asking questions is not a reason to > change the language - there always will be people asking questions, and > that's fine as long as we have good easily accessible answers, which we do. In Germany we say "jain" (yes and no). You are right that you brought up that argument and I acknowledged it but put it like you did not bring up one in my last message. I am sorry for that. You are not right that Stackoverflow is my only argument. I provided scientific proof that those aliases harm the design of a language and a more general proof in regards to usability and duplication. People often rely on side effects but that does not mean that this is a good practice. We are talking here about a feature that was added in PHP 4 in order to support OO coding. The feature is not used anymore by a major part of the community and it is considered best practice to use the proper access modifiers (namely public in this case). We are talking about a future version of PHP (probably 8) and not about removing it from PHP 4, 5, nor 7. There is only one constant in life and that is change. Insisting on not changing means to stop; especially in a context of software development and engineering (which actually implies constant change in its own name). -- Richard "Fleshgrinder" Fussenegger --- ## #91391 — Tony Marston — 2016-02-25T09:26:20+00:00 URL: https://externals.io/message/91330#91391 · In reply to: #91379 wrote in message news:56CDEB49.5040006@fleshgrinder.com... > > > > […] > >Hey there. :) > > > > […] > >My example seems to be bad or to abstract. I am not thinking that a >programming language can become that intuitive; the contrary. >Programming languages are already so complicated that it is harmful to >start littering them with thousands of ways to achieve one thing. It >also makes users uncertain when the manual entry says that something is >an alias. They fear that it might be removed. That fear is unjustified. They need to learn the difference between "duplicated" and "deprecated". Unless a function is marked as "deprecated" it cannot be removed. >They create project rules that forbid the usage of aliases ... We cannot stop stupid people from creating stupid rules. >Science shows that it is harmful, let's clean it up! Your "proof" is not scientific, it is just personal opinion. There is no evidence that use of the "var" keyword is harmful in any way. > > […] > >In Germany we say "jain" (yes and no). You are right that you brought up >that argument and I acknowledged it but put it like you did not bring up >one in my last message. I am sorry for that. You are not right that >Stackoverflow is my only argument. I provided scientific proof that >those aliases harm the design of a language and a more general proof in >regards to usability and duplication. Again, your "proof" is NOT scientific, it is merely a personal opinion. >People often rely on side effects but that does not mean that this is a >good practice. We are talking here about a feature that was added in PHP >4 in order to support OO coding. The feature is not used anymore by a >major part of the community and it is considered best practice to use Where is your proof? You say "not used by a major part of the community" which means that it is still being used by a minor part, but exactly how "minor"? I don't see why I should be forced to make a totally unnecessary change to vast numbers of my scripts just to fall in line with your personal opinions. >the proper access modifiers (namely public in this case). We are talking >about a future version of PHP (probably 8) and not about removing it >from PHP 4, 5, nor 7. There is no reason to remove it from ANY version of PHP. It does no harm, it would take effort to take it out and amend the documentation, but for what benefit? >There is only one constant in life and that is change. Insisting on not >changing means to stop; especially in a context of software development >and engineering (which actually implies constant change in its own name). Change for change's sake is never a good idea. I have been developing in several languages for 40 years, and I can tell you point blank that while programmers expect new features to be added they do NOT expect old features to disappear. Once a piece of code has been written and has proved to work as designed it is expected to work with all future versions. The only exception to this is to plug holes in security. This is called "forwards compatibility", and was a major selling point of all my previous languages. If developers fear that they will have to rewrite huge swathes of code each time a new version is released they will quickly give up and move to a "professional" language which offers long term stability. -- Tony Marston --- ## #91412 — Fleshgrinder — 2016-02-25T18:10:39+00:00 URL: https://externals.io/message/91330#91412 · In reply to: #91391 > > […] > > Your "proof" is not scientific, it is just personal opinion. There is no > evidence that use of the "var" keyword is harmful in any way. I think the diverged from talking about the "var" keyword in particular towards duplication in general a long time ago. However, I still think that DRY is empirically proven. > Where is your proof? You say "not used by a major part of the community" > which means that it is still being used by a minor part, but exactly how > "minor"? I don't see why I should be forced to make a totally > unnecessary change to vast numbers of my scripts just to fall in line > with your personal opinions. It is true that I did not provide this proof because Colin O'Dell did claim this fact in the very initial message of this thread and I believe him. > There is no reason to remove it from ANY version of PHP. It does no > harm, it would take effort to take it out and amend the documentation, > but for what benefit? I stick to the main reason I gave, DRY. Duplication needs to be managed and removing it removes the maintenance burden. > Change for change's sake is never a good idea. I have been developing in > several languages for 40 years, and I can tell you point blank that > while programmers expect new features to be added they do NOT expect old > features to disappear. Once a piece of code has been written and has > proved to work as designed it is expected to work with all future > versions. The only exception to this is to plug holes in security. This > is called "forwards compatibility", and was a major selling point of all > my previous languages. If developers fear that they will have to rewrite > huge swathes of code each time a new version is released they will > quickly give up and move to a "professional" language which offers long > term stability. I did not say that we should change for change's sake. I only stated that trying hard to prevent change by all means is wrong. Again, this diverged away from the "var" keyword alone a long time ago and was more a general statement. TL;DR Thanks for yet another aggressive/provocative email, I stick to my +1. However, you all have valid points to keep it. -- Richard "Fleshgrinder" Fussenegger --- ## #91423 — Yasuo Ohgaki — 2016-02-25T21:30:56+00:00 URL: https://externals.io/message/91330#91423 · In reply to: #91412 Hi all, > > > […] > > I think the diverged from talking about the "var" keyword in particular > towards duplication in general a long time ago. However, I still think > that DRY is empirically proven. I'm 0 for this change. Although it's close to +1. Those who are willing to remove "var" from PHP should write conversion script that scans PHP scripts and converts "var" to "public". Then there will be more supporters for removing "var". We have tokenizer. It should not be difficult. Tokenizier may be extended to make this kind of conversion script easier. i.e. Get offset of tokens also. If there is offset, writing conversion script is trivial. It seems some work is required to return offset, though. Regards, -- Yasuo Ohgaki yohgaki@ohgaki.net --- ## #91429 — Tony Marston — 2016-02-26T09:32:47+00:00 URL: https://externals.io/message/91330#91429 · In reply to: #91412 wrote in message news:56CF439F.2040506@fleshgrinder.com... > > > > […] > >I think the diverged from talking about the "var" keyword in particular >towards duplication in general a long time ago. However, I still think >that DRY is empirically proven. > > > […] > >It is true that I did not provide this proof because Colin O'Dell did >claim this fact in the very initial message of this thread and I believe >him. So just because someone else states an opinion you are prepared to convert this to a "fact"? > > […] > >I stick to the main reason I gave, DRY. Duplication needs to be managed >and removing it removes the maintenance burden. What maintenance burden? How much time is taken up by the language developers in maintaining the code that deals with the "var" keyword? How many bugs have been reported on this word since PHP was released? > > […] > >I did not say that we should change for change's sake. I only stated >that trying hard to prevent change by all means is wrong. You are contradicting yourself. On the one hand you are saying that change for change's sake is wrong, and on the other hand you are saying that we should not block change for any reason. >Again, this >diverged away from the "var" keyword alone a long time ago and was more >a general statement. > >TL;DR Thanks for yet another aggressive/provocative email, I stick to my >+1. However, you all have valid points to keep it. > -- Tony Marston