Re: [sage-devel] Re: Should we still support old-style packages?

Hi, On Thu, Sep 10, 2015 at 04:42:32PM +0200, Jeroen Demeyer wrote: > On 2015-09-10 15:14, Nathann Cohen wrote: > >I do not have your faith that *all* of the guys who appear as > >maintainers of a package would, when receiving such a bug report, > >immediately fix it or make sure that it will not

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-12 09:48, Simon King wrote: All versions from 2.1.2 till now were at least partially motivated by coping with backward incompatible changes. I am probably responsible for several of these backward incompatible changes. However, they were changes to the basic Sage infrastructure,

[sage-devel] Re: Should we still support old-style packages?

Hi John, On 2015-09-12, John H Palmieri wrote: > By the way, I think the old database is broken: I think it tries to install > into some directory which now no longer exists. So I think the one in the > newest beta is the only option. In the same way (changed

[sage-devel] Re: Should we still support old-style packages?

On Saturday, September 12, 2015 at 9:48:38 AM UTC+2, Simon King wrote: > > For the record: I think ANY backward incompatible internal change that > *can* > be covered by a deprecation warning (thus, changed import locations, > changed function names or argspecs, etc) *has* to be covered by a >

[sage-devel] Re: Should we still support old-style packages?

> > I thought some patchbots did test optional packages on a more or less >> regular >> basis? >> > > No, the build/patchbots only test standard packages. Of course we would be > happy to have you volunteer as the person in change of testing all optional > packages. > > R has the resources

Re: [sage-devel] Re: Should we still support old-style packages?

> The new thing where all installed optional > packages are tested is a good start, it's clearly already helped. Here is a (stalled) attempt at pushing it further. Needs volunteers. http://trac.sagemath.org/ticket/18904 Nathann -- You received this message because you are subscribed to the

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-10, John Cremona wrote: > Am I right that Simon's package is one (perhaps the only one) for which (B) > would be useful? If that is true then I don't wish to make things hard for > Simon. But then I don't know what the obstacles are to converting his spkg.

Re: [sage-devel] Re: Should we still support old-style packages?

On 10 September 2015 at 14:02, Simon King wrote: > On 2015-09-10, John Cremona wrote: > > Am I right that Simon's package is one (perhaps the only one) for which > (B) > > would be useful? If that is true then I don't wish to make things hard >

Re: [sage-devel] Re: Should we still support old-style packages?

Hello, > Something else: Of course each package has a maintainer resp. an > upstream contact. Can this information be automatically extracted? Err. Well, "not always". Yesterday Dima converted a .spkg to new-style, but that package did not include a SPKG.txt file (so no official maintainer

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-10 15:14, Nathann Cohen wrote: I do not have your faith that *all* of the guys who appear as maintainers of a package would, when receiving such a bug report, immediately fix it or make sure that it will not be forgotten. Asking people to write to sage-devel sounds safer to me. +1

[sage-devel] Re: Should we still support old-style packages?

Hi John, On 2015-09-10, John Cremona wrote: > Has anyone started? AFter reading the above I just downloaded MeatAxe > 2.4.24 (2011) and very easily built it (make check OK); all one has to do > is copy Makefile to Makefile.conf and edit 2 lines, as I am sure you know.

Re: [sage-devel] Re: Should we still support old-style packages?

Re (B): a deprecation warning is normally a message to the user that they should start to do things in a different way, as the old way will stop working. This does not seem appropriate here, since it is not the user who is expected to change behaviour (unless you regard all users as developers!).

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 10:50, John Cremona wrote: Re (B): a deprecation warning is normally a message to the user that they should start to do things in a different way, as the old way will stop working. This does not seem appropriate here, since it is not the user who is expected to change behaviour

[sage-devel] Re: Should we still support old-style packages?

IMHO showing a warning that you are about to do something stupid but then proceeding to do so is a terrible user interface. Also, I don't like command line utilities that ask stupid questions (a.k.a. the windows UI style). Really, I still don't know a use case where you actually want to search

[sage-devel] Re: Should we still support old-style packages?

Another way to "fix" the problem: (E) Convert the remaining old-style packages to new-style packages. Nathann -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 10:15, Nathann Cohen wrote: Another way to "fix" the problem: (E) Convert the remaining old-style packages to new-style packages. Of course we should do that, but that's orthogonal to the question I asked. I don't think that (E) is going to happen short-term. As long as there

Re: [sage-devel] Re: Should we still support old-style packages?

> Of course we should do that, but that's orthogonal to the question I asked. > I don't think that (E) is going to happen short-term. > > As long as there is still at least one old-style package, we need to define > what should happen when a user types I suspect that it is not so orthogonal, and

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 11:56:28 AM UTC+2, Jeroen Demeyer wrote: > > But even if there is just a single old-style package > which works and is useful > I disagree with that rationale. We want to maximize value for the scientific community. If that means inconveniencing a single user

Re: [sage-devel] Re: Should we still support old-style packages?

On 9 September 2015 at 10:22, Jeroen Demeyer wrote: > On 2015-09-09 10:50, John Cremona wrote: > >> Re (B): a deprecation warning is normally a message to the user that >> they should start to do things in a different way, as the old way will >> stop working. This does

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 13:14, Volker Braun wrote: Because not even 1 out of 100 will figure out that "python" actually tries to install the old and unsupported package. There are other solutions to that problem, any of the following will do: * a proper implementation of (B) to make them figure it out *

Re: [sage-devel] Re: Should we still support old-style packages?

>> This way they would all become new-style packages at a very low cost. > > And which problem would that solve? It would then be possible to install them with "sage -i ", which is I belive the reason you created this thread? Nathann -- You received this message because you are subscribed to

Re: [sage-devel] Re: Should we still support old-style packages?

Another option: would it be possible to make "fake new-style packages", whose upstream/ tarball would be the .spkg? This way they would all become new-style packages at a very low cost. Nathann On 9 September 2015 at 10:15, Nathann Cohen wrote: > Another way to "fix"

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 11:26, John Cremona wrote: Not quite -- I was proposing to display the message and *not* install the package OK, so this is (D'): sage -i OLDSTYLEPKGNAME should give an error. However, it should determine whether an old-style package named OLDSTYLEPKGNAME exists and adjust the

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 12:01, Volker Braun wrote: in order to make Sage accessible to 100 other users then we should do that. How does option (D) make "Sage accessible to 100 other users" compared with options (B) or (C)? -- You received this message because you are subscribed to the Google Groups

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 13:33, Nathann Cohen wrote: Another option: would it be possible to make "fake new-style packages", whose upstream/ tarball would be the .spkg? This way they would all become new-style packages at a very low cost. And which problem would that solve? -- You received this message

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 12:13:39 PM UTC+2, Jeroen Demeyer wrote: > How does option (D) make "Sage accessible to 100 other users" compared > with options (B) or (C)? > Because not even 1 out of 100 will figure out that "python" actually tries to install the old and unsupported

Re: [sage-devel] Re: Should we still support old-style packages?

> But by reverting #19004 (which is essentially what #19158 does), it is also > possible to install them using "sage -i ". There are multiple ways out indeed. All I try to do here is help you two work together if I can, by suggesting "middle ways". What happens with unmaintained packages does not

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 14:34, Nathann Cohen wrote: This way they would all become new-style packages at a very low cost. And which problem would that solve? It would then be possible to install them with "sage -i ", which is I belive the reason you created this thread? But by reverting #19004

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 2:14:23 PM UTC+2, Jeroen Demeyer wrote: > > There are other solutions to that problem, any of the following will do: > * a proper implementation of (B) to make them figure it out > * the proposed #19105 > * removing the old python spkg from the server > None

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 2:47:20 PM UTC+2, Jeroen Demeyer wrote: > > But by reverting #19004 (which is essentially what #19158 does), it is > also possible to install them using "sage -i ". It is possible to try, and almost always fail miserably, to install them. -- You received

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 15:06, Volker Braun wrote: Other than deleting them all from the server, of course. The latter *would* also solve the problem. Just like reformatting your hard drive will instantly remove all software issues that you might have with your computer. -- You received this message

[sage-devel] Re: Should we still support old-style packages?

Hi Jeroen, On 2015-09-09, Jeroen Demeyer wrote: > The question is: to what extent should we continue supporting old-style > packages? > > (A) sage -i OLDSTYLEPKGNAME should just install the package, there is no > difference with new-style packages from the user's point

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Nathann Cohen wrote: > Another way to "fix" the problem: > (E) Convert the remaining old-style packages to new-style packages. Sure, but that means to teach noobs like me to use github in order to artificially create an "upstream" source (basically the

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 4:43:47 PM UTC+2, kcrisman wrote: > > TOPCOM vs. topcom seems very unlikely to cause a Sage installation to > self-destruct to me > It doesn't cause data loss but it is confusing enough that even experienced developers can't tell which one to use (see

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Volker Braun wrote: > I disagree with that rationale. We want to maximize value for the > scientific community. If that means inconveniencing a single user in order > to make Sage accessible to 100 other users then we should do that. Are you saying that

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Nathann Cohen wrote: > I suspect that it is not so orthogonal, and that it is the main source > of disagreement between Volker and you. He seems to say that "all > packages which work are now new-style", so that there is no "useful" > old-style package

[sage-devel] Re: Should we still support old-style packages?

Hi John, On 2015-09-09, John Cremona wrote: > Not quite -- I was proposing to display the message and *not* install the > package -- since I thought that essentially none of the old .spkg will now > install anyway? That old .spkg do not install is not just a fact, but in

[sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 5:30:54 PM UTC+2, Simon King wrote: > > For the record: A deprecation warning does *not* mean that the user is > about to do something stupid, but that (s)he is doing something that used > to be supported but soon will not be. > Exactly, which is why a

[sage-devel] Re: Should we still support old-style packages?

Hi Volker, On 2015-09-09, Volker Braun wrote: > On Wednesday, September 9, 2015 at 5:45:14 PM UTC+2, Simon King wrote: >> >> However, I believe that "sage -i " should >> still be supported. > > > Agree, and this is not what this thread is about. Is it not? OK, then I

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Jeroen Demeyer wrote: > On 2015-09-09 13:14, Volker Braun wrote: >> Because not even 1 out of 100 will figure out that "python" actually >> tries to install the old and unsupported package. > > There are other solutions to that problem, any of the following

[sage-devel] Re: Should we still support old-style packages?

Hi Nathann, On 2015-09-09, Nathann Cohen wrote: >> But by reverting #19004 (which is essentially what #19158 does), it is also >> possible to install them using "sage -i ". > > There are multiple ways out indeed. All I try to do here is help you > two work together if I

[sage-devel] Re: Should we still support old-style packages?

Hi Travis, On 2015-09-09, Travis Scrimshaw wrote: >To the discussion at hand, I think the current system of not allowing > old spkgs to be installed is a waste. What are the current old style > (non-experimental) spkgs, which of them have you tried, and which of them >

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Volker Braun wrote: > IMHO showing a warning that you are about to do something stupid but then > proceeding to do so is a terrible user interface. Also, I don't like > command line utilities that ask stupid questions (a.k.a. the windows UI > style). Are

[sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 5:45:14 PM UTC+2, Simon King wrote: > > However, I believe that "sage -i " should > still be supported. Agree, and this is not what this thread is about. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To

Re: [sage-devel] Re: Should we still support old-style packages?

A possible (hack) fix is to put something in the skpg script to explicitly ban it from installing python without specifying 2 or 3. It keeps the old python spkg for backwards compatibility for 5.x and prevents 6.x users from using the old outdated version. Either that or we provide a

[sage-devel] Re: Should we still support old-style packages?

I think that we should try to support old-style packages. Given that many of them are broken, and given that there are only 30 optional old-style packages (at least on http://files.sagemath.org/spkg/optional/), we can just look at them and white-list the ones that work and/or are not ancient.

Re: [sage-devel] Re: Should we still support old-style packages?

> But by reverting #19004 (which is essentially what #19158 does), it is >> also possible to install them using "sage -i ". > > > It is possible to try, and almost always fail miserably, to install them. > It would be helpful for this discussion to know what packages we are actually talking

Re: [sage-devel] Re: Should we still support old-style packages?

> >To the discussion at hand, I think the current system of not allowing > old spkgs to be installed is a waste. What are the current old style > (non-experimental) spkgs, which of them have you tried, and which of them > don't work on what system. I'm willing to put in some time and

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 16:42, Travis Scrimshaw wrote: A possible (hack) fix is to put something in the skpg script to explicitly ban it from installing python without specifying 2 or 3. Assuming #19119, it's very easy to this. Just two short lines need to be added to build/make/deps (one for sage -i

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Volker Braun wrote: > On Wednesday, September 9, 2015 at 5:30:54 PM UTC+2, Simon King wrote: >> >> For the record: A deprecation warning does *not* mean that the user is >> about to do something stupid, but that (s)he is doing something that used >> to be

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 1:24:09 PM UTC-7, Jeroen Demeyer wrote: > > On 2015-09-09 22:17, Thierry wrote: > > sorry for not participating actively to this long thread, i am currently > > pretty far from computers. Let me just mention that this discussion > > already appears some time

[sage-devel] Re: Should we still support old-style packages?

LiE is ready at http://trac.sagemath.org/ticket/19176. cluster_seed is next on http://trac.sagemath.org/ticket/19177. coxeter3 is after that on http://trac.sagemath.org/ticket/19178. Best, Travis On Wednesday,

Re: [sage-devel] Re: Should we still support old-style packages?

Thank you Jeroen and John for posting those links. coxeter3 and lie are useful to me and are things I'd want to keep and I will convert over. I will also be handling cluster_seed. Best, Travis On Wednesday, September 9, 2015 at 12:25:53 PM UTC-5, John H Palmieri wrote: > > > > On Wednesday,

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 18:36, Dima Pasechnik wrote: compared to the effort of keeping the whole support for them... There is no "effort" needed, we just need to undo the intentional breaking of the support. -- You received this message because you are subscribed to the Google Groups "sage-devel"

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 19:14, Nathann Cohen wrote: Is there a list of those old spkg somewhere? http://files.sagemath.org/spkg/index.html We could see what there is to do by putting it on a wiki page Like this one: http://wiki.sagemath.org/Classify%20old-style%20packages -- You received this message

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 10:09:21 AM UTC-7, Jeroen Demeyer wrote: > > On 2015-09-09 17:06, John H Palmieri wrote: > > I think that we should white-list at least some of these, and also try > > to convert some of them (all of them?) to new-style packages. Same with > > Simon's

Re: [sage-devel] Re: Should we still support old-style packages?

See http://trac.sagemath.org/ticket/19175 for ore_algebra. On Wednesday, September 9, 2015 at 11:38:34 AM UTC-7, Travis Scrimshaw wrote: > > Thank you Jeroen and John for posting those links. > > coxeter3 and lie are useful to me and are things I'd want to keep and I > will convert over. I

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, John H Palmieri wrote: > See http://trac.sagemath.org/ticket/19175 for ore_algebra. See http://trac.sagemath.org/ticket/18514 for p_group_cohomology. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To

Re: [sage-devel] Re: Should we still support old-style packages?

On Wed, Sep 09, 2015 at 01:20:26PM -0700, kcrisman wrote: > Is biopython pip-installable? Yes. If you want to test without changing your Sage install, just install virtualenv and do: virtualenv venv . venv/bin/activate pip install biopython # ipython # 

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Volker Braun wrote: > On Wednesday, September 9, 2015 at 9:39:49 PM UTC+2, Jeroen Demeyer wrote: >> >> But this argument completely fails when backwards compatibility is >> involved. Actively removing a feature (even if broken 99% of the time) >> without

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, 9 September 2015 03:01:53 UTC-7, Volker Braun wrote: > > On Wednesday, September 9, 2015 at 11:56:28 AM UTC+2, Jeroen Demeyer wrote: >> >> But even if there is just a single old-style package >> which works and is useful >> > > I disagree with that rationale. We want to maximize

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 10:14:44 AM UTC-7, Nathann Cohen wrote: > > > That's of course a noble goal, but it doesn't answer the question about > what > > to do on the short term (unless you volunteer to port all packages > before > > the next Sage release). > > Is there a list of

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 18:24, Simon King wrote: Hi Volker, On 2015-09-09, Volker Braun wrote: On Wednesday, September 9, 2015 at 5:45:14 PM UTC+2, Simon King wrote: However, I believe that "sage -i " should still be supported. Agree, and this is not what this thread is

Re: [sage-devel] Re: Should we still support old-style packages?

> That's of course a noble goal, but it doesn't answer the question about what > to do on the short term (unless you volunteer to port all packages before > the next Sage release). Is there a list of those old spkg somewhere? We could see what there is to do by putting it on a wiki page and

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, 9 September 2015 10:25:53 UTC-7, John H Palmieri wrote: > > > > On Wednesday, September 9, 2015 at 10:09:21 AM UTC-7, Jeroen Demeyer wrote: >> >> On 2015-09-09 17:06, John H Palmieri wrote: >> > I think that we should white-list at least some of these, and also try >> > to

Re: [sage-devel] Re: Should we still support old-style packages?

> if there is just one old-style package left, then the effort to convert it > into new style is small, compared to the effort of > keeping the whole support for them... +1 Let us remove those that are not supposed to work (like the old version of Simon's) and see what remains. Perhaps there

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 17:06, John H Palmieri wrote: I think that we should white-list at least some of these, and also try to convert some of them (all of them?) to new-style packages. Same with Simon's p_group_cohomology package, which didn't build correctly for me. I don't have objections to leaving

Re: [sage-devel] Re: Should we still support old-style packages?

On Wed, Sep 9, 2015 at 8:22 AM, Simon King wrote: > On 2015-09-09, Nathann Cohen wrote: >> Another way to "fix" the problem: >> (E) Convert the remaining old-style packages to new-style packages. > I personally really like this suggestion. >

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 19:03, William Stein wrote: On Wed, Sep 9, 2015 at 8:22 AM, Simon King wrote: On 2015-09-09, Nathann Cohen wrote: Another way to "fix" the problem: (E) Convert the remaining old-style packages to new-style packages. I

Re: [sage-devel] Re: Should we still support old-style packages?

Hi, sorry for not participating actively to this long thread, i am currently pretty far from computers. Let me just mention that this discussion already appears some time ago, and that somehow, the mailing-list is probably not the appropriate tool for solving such an issue, while the wiki is

Re: [sage-devel] Re: Should we still support old-style packages?

> > > instead of reading and posting here, I just spent time converting > jones_numfield to new style. (#19174) > If every poster here does spend half a day or a day on this, the issue > will be moot... :-) > > > :) Is biopython pip-installable? -- You received this message because you are

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 22:17, Thierry wrote: sorry for not participating actively to this long thread, i am currently pretty far from computers. Let me just mention that this discussion already appears some time ago, and that somehow, the mailing-list is probably not the appropriate tool for solving such

Re: [sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 9:39:49 PM UTC+2, Jeroen Demeyer wrote: > > But this argument completely fails when backwards compatibility is > involved. Actively removing a feature (even if broken 99% of the time) > without deprecation is a very exceptional move which should require >

[sage-devel] Re: Should we still support old-style packages?

On Wednesday, 9 September 2015 11:59:04 UTC-7, Simon King wrote: > > On 2015-09-09, John H Palmieri wrote: > > See http://trac.sagemath.org/ticket/19175 for ore_algebra. > > See http://trac.sagemath.org/ticket/18514 for p_group_cohomology. > the links to the spkg no

[sage-devel] Re: Should we still support old-style packages?

On Wednesday, September 9, 2015 at 6:00:22 PM UTC+2, Simon King wrote: > > Are you saying that the fact that something *works* > If FEATURE gives you 1 case that works and 99 that fail then it is not a good feature. Just working rarely is not good enough. -- You received this message because

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Dima Pasechnik wrote: >> See http://trac.sagemath.org/ticket/18514 for p_group_cohomology. >> > > the links to the spkg no longer work there, please update That has been discussed here already. Since sage.math.washington doesn't allow access, I need to find a

Re: [sage-devel] Re: Should we still support old-style packages?

On 2015-09-09 21:34, Volker Braun wrote: On Wednesday, September 9, 2015 at 6:00:22 PM UTC+2, Simon King wrote: Are you saying that the fact that something *works* If FEATURE gives you 1 case that works and 99 that fail then it is not a good feature. Just working rarely is not good

[sage-devel] Re: Should we still support old-style packages?

On 2015-09-09, Simon King wrote: > That has been discussed here already. Since sage.math.washington doesn't > allow access, I need to find a different place. The spkg is too large > for an attachment to trac, but it seems that the university of Jena is > willing to host