DVCSes, bug trackers, and GNUness

DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

So Emacs must use bzr because bzr is GNU, fine. I'm all for GNU standing united.

But Emacs is using debbugs for bug tracking instead of the Savannah
bug trackers, because Savannah doesn't have an email interface for
manipulating bugs.

Shouldn't we switch to Savannah and fix it? Isn't that what we're
supposed to do about bzr? Why isn't there an executive decision to
prefer the GNU package?

I'm saying this because in Octave, another GNU package, we do use
Savannah, and boy would I like to have it fixed. I'm working on doing
just that by finishing the Python rewrite of Savane, and if email
support is what it takes to stop using debbugs, I would very much like
some help doing that.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Miles Bader]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
> Shouldn't we switch to Savannah and fix it? Isn't that what we're
> supposed to do about bzr? Why isn't there an executive decision to
> prefer the GNU package?

Er, well you still need somebody to actually implement the feature.
An "executive decision" seems pointless absent actual code....

[So a better plan would seem to be:  _first_ find somebody to do the
work, _then_ once things seem in better shape, worry about ideological
purity...]

-miles

-- 
Selfish, adj. Devoid of consideration for the selfishness of others.

Re: DVCSes, bug trackers, and GNUness

[Richard Stallman]

    Shouldn't we switch to Savannah and fix it? Isn't that what we're
    supposed to do about bzr? Why isn't there an executive decision to
    prefer the GNU package?

Savannah isn't a package, it's a facility.  So it is a different kind
of issue.

However, it would be very good to work on improving Savannah.
Please do, if you want to.

--
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call

Re: DVCSes, bug trackers, and GNUness

[Chong Yidong]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:

> in Octave, another GNU package, we do use Savannah, and boy would I
> like to have it fixed. I'm working on doing just that by finishing the
> Python rewrite of Savane, and if email support is what it takes to
> stop using debbugs, I would very much like some help doing that.

Thanks for helping to work on Savane.

One nice thing about debbugs.gnu.org, though, is that it's a dedicated
VM that we (i.e. some Emacs developers) have root access to.  So we can
tweak stuff without worrying about breaking thousands of dependent
projects.  At this point, the system works smoothly enough that it's not
worth migrating away.

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/24 Miles Bader <miles@gnu.org>:
> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>> Shouldn't we switch to Savannah and fix it? Isn't that what we're
>> supposed to do about bzr? Why isn't there an executive decision to
>> prefer the GNU package?
>
> Er, well you still need somebody to actually implement the feature.
> An "executive decision" seems pointless absent actual code....

Nah, isn't that why we're using bzr even though most people hate it?
Not because it's good code, but because it's GNU? So the executive
decision there worked.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Richard Stallman <rms@gnu.org>:
>    Shouldn't we switch to Savannah and fix it? Isn't that what we're
>    supposed to do about bzr? Why isn't there an executive decision to
>    prefer the GNU package?
>
> Savannah isn't a package, it's a facility.  So it is a different kind
> of issue.

I don't understand the distinction. Can you explain?

> However, it would be very good to work on improving Savannah.
> Please do, if you want to.

I do, but I would like to coerce other people to help me, just like
people are being coerced right now to work with bzr and hopefully
improve it.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Lars Magne Ingebrigtsen]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:

> Shouldn't we switch to Savannah and fix it? Isn't that what we're
> supposed to do about bzr? Why isn't there an executive decision to
> prefer the GNU package?

bzr and git offer the same feature set.  The practical differences
between bzr and git are minuscule.

The practical differences between debbugs and trackers like the Savannah
one are rather huge.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: DVCSes, bug trackers, and GNUness

[Juanma Barranquero]

2012/4/25 Jordi Gutiérrez Hermoso <jordigh@octave.org>:

> Nah, isn't that why we're using bzr even though most people hate it?

Most people hate it? Really?

    Juanma

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Lars Magne Ingebrigtsen <larsi@gnus.org>:
> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>
>> Shouldn't we switch to Savannah and fix it? Isn't that what we're
>> supposed to do about bzr? Why isn't there an executive decision to
>> prefer the GNU package?
>
> bzr and git offer the same feature set.  The practical differences
> between bzr and git are minuscule.

Y'know, it's a false dichotomy between bzr and git... It's not as if
git is the only other option. I think hg is far more usable than git
and does not have the performance disadvantage that bzr presents.

At any rate, I'm sure you will find no shortage of opinions from
people who think git is far more featureful than bzr.

> The practical differences between debbugs and trackers like the Savannah
> one are rather huge.

The same can be said between bzr and another DVCS. Really, the *one*
feature that debbugs has and Savane (the software that Savannah runs)
doesn't is an email interrface.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Juanma Barranquero <lekktu@gmail.com>:
> 2012/4/25 Jordi Gutiérrez Hermoso <jordigh@octave.org>:
>
>> Nah, isn't that why we're using bzr even though most people hate it?
>
> Most people hate it? Really?

Yeah, look around you. Do I really need to link to the endless
flamewars about why Emacs isn't using git?

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Miles Bader]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>> Er, well you still need somebody to actually implement the feature.
>> An "executive decision" seems pointless absent actual code....
>
> Nah, isn't that why we're using bzr even though most people hate it?
> Not because it's good code, but because it's GNU? So the executive
> decision there worked.

However the "executive decision" in that case came _after_ the
presence of a usable[*] tool.

[*] er, well...

-Miles

-- 
o The existentialist, not having a pillow, goes everywhere with the book by
  Sullivan, _I am going to spit on your graves_.

Re: DVCSes, bug trackers, and GNUness

[Lars Magne Ingebrigtsen]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:

> The same can be said between bzr and another DVCS. Really, the *one*
> feature that debbugs has and Savane (the software that Savannah runs)
> doesn't is an email interrface.

Does Savane have an API?

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: DVCSes, bug trackers, and GNUness

[Juanma Barranquero]

2012/4/25 Jordi Gutiérrez Hermoso <jordigh@octave.org>:

> Yeah, look around you. Do I really need to link to the endless
> flamewars about why Emacs isn't using git?

Oh, you mean that the git fanboys are angry. That's quite different
from "most people hate it".

For the purposes of this discussion, that I hope goes nowhere *fast*,
I define "git fanboys" as people in one of these two categories: a)
those that would consider anything by Linus Torvalds as a gift from
heaven; b) those whose argument (really, they only have one) to prefer
git is that it is "fast". People with a more nuanced vision of dVCS
are excluded from my definition and need not feel offended by it.

    Juanma

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Lars Magne Ingebrigtsen <larsi@gnus.org>:
> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>
>> The same can be said between bzr and another DVCS. Really, the *one*
>> feature that debbugs has and Savane (the software that Savannah runs)
>> doesn't is an email interrface.
>
> Does Savane have an API?

What does that mean? Also, *which* Savane? Sorry I wasn't clear
earlier. The current one is PHP and probably doesn't have anything
that you would consider an API. The rewrite I'm trying to revive,

    http://inversethought.com/hg/savane-forge

is in Python and uses Django. Look, it's not difficult to get it up
and running in a local test environment:

    http://jordi.platinum.linux.pl/piccies/savane-screenshot-1.png

Since Django is a commonly used web, uh, thing (framework? platform?),
whatever API someone has written for Django is available for the
Savane rewrite.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Juanma Barranquero <lekktu@gmail.com>:
> 2012/4/25 Jordi Gutiérrez Hermoso <jordigh@octave.org>:
>
>> Yeah, look around you. Do I really need to link to the endless
>> flamewars about why Emacs isn't using git?
>
> Oh, you mean that the git fanboys are angry. That's quite different
> from "most people hate it".

Look, you don't need to convince me that git sucks and its proponents
are overzealous:

    http://jordi.inversethought.com/blog/on-gitology/

However, it's clear they vastly outnumber you and me. We're not using
bzr because it's the popular choice. We're using it because it's GNU
and a small minority of people have pushed this decision on everyone
else.

So can we push a similar decision for the BTS? debbugs isn't GNU.
Savane is GNU. Other GNU packages use Savane, so why doesn't Emacs?

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Lars Magne Ingebrigtsen]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:

>> Does Savane have an API?
>
> What does that mean?

Like...  an API.  :-)

We(*) don't use the debbugs web interface for looking at bug reports, you
know.   `M-x debbugs-gnu'.

----
*) Somewhat royal.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Lars Magne Ingebrigtsen <larsi@gnus.org>:
> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>
>>> Does Savane have an API?
>>
>> What does that mean?
>
> Like...  an API.  :-)

No, really, I want to convince you to stop using debbugs. Please
explain to me what this means. You want a network protocol that Emacs
can use to query the status of the BTS?

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Juanma Barranquero]

2012/4/25 Jordi Gutiérrez Hermoso <jordigh@octave.org>:

> Look, you don't need to convince me that git sucks and its proponents
> are overzealous:

I'm quite glad not to count you among the git fanboys. Congratulations.

> We're not using
> bzr because it's the popular choice. We're using it because it's GNU
> and a small minority of people have pushed this decision on everyone
> else.

Which is true, and it's called "politics". That's quite different from
your previous claim that "most people hate [bzr]".

    Juanma

Re: DVCSes, bug trackers, and GNUness

[Michael Albinus]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:

> 2012/4/25 Lars Magne Ingebrigtsen <larsi@gnus.org>:
>> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>>
>>>> Does Savane have an API?
>>>
>>> What does that mean?
>>
>> Like...  an API.  :-)
>
> No, really, I want to convince you to stop using debbugs. Please
> explain to me what this means. You want a network protocol that Emacs
> can use to query the status of the BTS?

Yes. BTS offers a SOAP interface, which we (not only His Royalness Lars)
use for querying bugs from inside Emacs.

> - Jordi G. H.

Best regards, Michael.

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Michael Albinus <michael.albinus@gmx.de>:
> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>
>> 2012/4/25 Lars Magne Ingebrigtsen <larsi@gnus.org>:
>>> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>>>
>>>>> Does Savane have an API?
>>>>
>>>> What does that mean?
>>>
>>> Like...  an API.  :-)
>>
>> No, really, I want to convince you to stop using debbugs. Please
>> explain to me what this means. You want a network protocol that Emacs
>> can use to query the status of the BTS?
>
> Yes. BTS offers a SOAP interface, which we (not only His Royalness Lars)
> use for querying bugs from inside Emacs.

So something like this?

    https://gist.github.com/499210

It really isn't so difficult to do. How much of this has to be implemented?

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Michael Albinus]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:

>> Yes. BTS offers a SOAP interface, which we (not only His Royalness Lars)
>> use for querying bugs from inside Emacs.
>
> So something like this?
>
>     https://gist.github.com/499210

Maybe. This code doesn't tell me too much. Is there a WSDL file, which
describes the SOAP interface?

> It really isn't so difficult to do. How much of this has to be implemented?

This question I don't understand. For BTS, we use the SOAP interface to
query for bugs using a mask of different bug attributes.  And we have
also a full text query.

BTS itself offers kind of Perl, Python, Ruby and PHP clients. For Emacs,
we have written our own client in Lisp.

> - Jordi G. H.

Best regards, Michael.

Re: DVCSes, bug trackers, and GNUness

[Jeremiah Dodds]

Juanma Barranquero <lekktu@gmail.com> writes:

> 2012/4/25 Jordi Gutiérrez Hermoso <jordigh@octave.org>:
>
>> Nah, isn't that why we're using bzr even though most people hate it?
>
> Most people hate it? Really?

I don't see much hate for it, other than the occasional very loud
person, but the majority of people I know with experience with the "big
name" VCSes, myself included consider it the "lesser" of the DVCSes, but
not in an incredibly major way.

That said, it's:

  A: A whole lot nicer than it was four years ago
  B: Obviously perfectly fine for emacs' purposes

If people really want to use their favorite system, bridges exist.

Re: DVCSes, bug trackers, and GNUness

[Stefan Monnier]

> Shouldn't we switch to Savannah and fix it?  Isn't that what we're
> supposed to do about bzr?  Why isn't there an executive decision to
> prefer the GNU package?

I'd rather not discuss the Bzr decision.  As for using a GNU bugtracker
instead of Debbugs, I don't have much of an opinion: Debbugs is far from
perfect, so from a technical point of view I'd be willing to use
something else, if it's better (the main weaknesses AFAICT of Debbugs
being its limited search&classification functionality, as well as its
fairly poor web interface).

Debbugs is Free (tho it seems to be GPLv2 only), so I'm perfectly happy
using it.  If a GNU package, such as Savane, were to provide a viable
replacement, I'd be happy to switch to it, tho the burden of moving over
would need to be justified by some additional benefit.

> Savannah, and boy would I like to have it fixed. I'm working on doing
> just that by finishing the Python rewrite of Savane, and if email
> support is what it takes to stop using debbugs, I would very much like
> some help doing that.

Actually, I think there are two issues that need to be addressed and
that tend to be bundled as "email support":
1- the ability to submit bug reports over email, to receive bug-reports
   via email, and to reply to bug-report via email.
2- the ability to manipulate the state of bug-reports (tag, close,
   merge, ...) offline.  In Debbugs this is done via email, and I guess
   some people would prefer if it stayed that way, but I think that many
   people would be happy to use another interface (e.g. something like
   `debbugs-gnu') for that.


        Stefan

Re: DVCSes, bug trackers, and GNUness

[Michael Albinus]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> 2- the ability to manipulate the state of bug-reports (tag, close,
>    merge, ...) offline.  In Debbugs this is done via email, and I guess
>    some people would prefer if it stayed that way, but I think that many
>    people would be happy to use another interface (e.g. something like
>    `debbugs-gnu') for that.

Btw, `debbugs-gnu' was written by Lars (and partially by me) in order to
satisfy our personal needs. If you have feature wishes, just say.

>         Stefan

Best regards, Michael.

Re: DVCSes, bug trackers, and GNUness

[Juanma Barranquero]

On Wed, Apr 25, 2012 at 19:05, Jeremiah Dodds <jeremiah.dodds@gmail.com> wrote:

>  A: A whole lot nicer than it was four years ago

That was in fact a secondary reason to choose it, IIRC. To push it to improve.

>  B: Obviously perfectly fine for emacs' purposes

Agreed.

    Juanma

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Stefan Monnier <monnier@iro.umontreal.ca>:
>> Shouldn't we switch to Savannah and fix it?  Isn't that what we're
>> supposed to do about bzr?  Why isn't there an executive decision to
>> prefer the GNU package?
>
> I'd rather not discuss the Bzr decision.

No, let's. Because I know you're tired of it, and I know why you are:
It's the decision everyone hated but taken only because of GNUness.

I want a similar decision for the BTS. If it's about GNUness, then
stop using debbugs. Technical merits don't matter, right? That's what
everyone said about bzr. So, if technical merits don't matter, I want
you to force everyone to stop using debbugs, no matter how painful it
is, and to start working with me on fixing Savane.

Be consistent. Either do it for all of Emacs infrastructure or none.
Don't be selective about what gets the GNU stamp of approval and what
doesn't.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Michael Albinus <michael.albinus@gmx.de>:
> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>
>>> Yes. BTS offers a SOAP interface, which we (not only His Royalness Lars)
>>> use for querying bugs from inside Emacs.
>>
>> So something like this?
>>
>>     https://gist.github.com/499210
>
> Maybe. This code doesn't tell me too much. Is there a WSDL file, which
> describes the SOAP interface?

It doesn't do much. Point is, doing SOAP with Python isn't difficult,
because other people have already built the necessary libraries to do
so.

>> It really isn't so difficult to do. How much of this has to be
>> implemented?
>
> This question I don't understand. For BTS, we use the SOAP interface
> to query for bugs using a mask of different bug attributes.  And we
> have also a full text query.
>
> BTS itself offers kind of Perl, Python, Ruby and PHP clients. For
> Emacs, we have written our own client in Lisp.

When you say "BTS" you mean "debbugs", right?

I suppose I should reimplement in Python the debbugs email interface
for discussing and manipulating bugs. I take it you use all of it, not
just a subset. I'll get working on this.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Karl Fogel]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>2012/4/25 Stefan Monnier <monnier@iro.umontreal.ca>:
>>> Shouldn't we switch to Savannah and fix it?  Isn't that what we're
>>> supposed to do about bzr?  Why isn't there an executive decision to
>>> prefer the GNU package?
>>
>> I'd rather not discuss the Bzr decision.
>
>No, let's. Because I know you're tired of it, and I know why you are:
>It's the decision everyone hated but taken only because of GNUness.
>
>I want a similar decision for the BTS. If it's about GNUness, then
>stop using debbugs. Technical merits don't matter, right? That's what
>everyone said about bzr. So, if technical merits don't matter, I want
>you to force everyone to stop using debbugs, no matter how painful it
>is, and to start working with me on fixing Savane.
>
>Be consistent. Either do it for all of Emacs infrastructure or none.
>Don't be selective about what gets the GNU stamp of approval and what
>doesn't.

That's not an accurate characterization of the Bzr decision.  This is
based just on my memory, but if you want to check this in the archives
I'm pretty sure they'll match up.

No one ever said that technical merits don't matter.  What several
people, including Richard, said was that all other things being equal --
or close enough to equal -- then Emacs should go with the GNU project.
That criterion was certainly met with Bzr.

If Bzr had required a heavy investment of development before it would be
meet the technical needs of the Emacs project, then the decision would
have been different.  But it already met the needs, so that issue didn't
come up.  (As it happens, it's gotten even better since then, so the
"all other things being equal" has slowly come closer to just "equal".)

I am not defending the decision to use Bzr.  I'm merely explaining that
it was made on a different basis than the one you claim above.  No
principle was in play that would imply we must also choose Savane now if
we want to be consistent.

-Karl

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Karl Fogel <kfogel@red-bean.com>:
> Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>>2012/4/25 Stefan Monnier <monnier@iro.umontreal.ca>:
>>>> Shouldn't we switch to Savannah and fix it?  Isn't that what we're
>>>> supposed to do about bzr?  Why isn't there an executive decision to
>>>> prefer the GNU package?
>>>
>>> I'd rather not discuss the Bzr decision.
>>
>>No, let's. Because I know you're tired of it, and I know why you are:
>>It's the decision everyone hated but taken only because of GNUness.
>>
>>I want a similar decision for the BTS. If it's about GNUness, then
>>stop using debbugs. Technical merits don't matter, right? That's
>>what everyone said about bzr. So, if technical merits don't matter,
>>I want you to force everyone to stop using debbugs, no matter how
>>painful it is, and to start working with me on fixing Savane.
>
> No one ever said that technical merits don't matter.  What several
> people, including Richard, said was that all other things being
> equal -- or close enough to equal -- then Emacs should go with the
> GNU project. That criterion was certainly met with Bzr.

Okay, so I'm getting a consistent message that if I want people to
prefer Savane over debbugs, I need to first make sure Savane can have
an email and SOAP interface. Is this accurate? Then it's a GNU project
with comparable features to debbugs, so then there will be no reason
to keep using debbugs. Am I understanding this correctly?

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Glenn Morris]

Jordi Gutiérrez Hermoso wrote:

> It's the decision everyone hated but taken only because of GNUness.

I didn't hate it, so you don't speak for me.
I hate the endlessly repeated, non-productive whining about it.

BTW:

http://savannah.nongnu.org/projects/savane-cleanup
"This project is not part of the GNU Project."

http://savannah.nongnu.org/p/savane
"This project is not part of the GNU Project."

http://www.gnu.org/software/gnats/
"The GNU bug tracking system"

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

2012/4/25 Glenn Morris <rgm@gnu.org>:
> Jordi Gutiérrez Hermoso wrote:
>
>> It's the decision everyone hated but taken only because of GNUness.
>
> I didn't hate it, so you don't speak for me.
> I hate the endlessly repeated, non-productive whining about it.
>
> BTW:
>
> http://savannah.nongnu.org/projects/savane-cleanup
> "This project is not part of the GNU Project."
>
> http://savannah.nongnu.org/p/savane
> "This project is not part of the GNU Project."

Yeah, but Savannah is hosted on gnu.org and runs Savane. If slapping
the GNU name on the project is all it takes (e.g. R and Gnome are also
"GNU", despite acting almost nothing like GNU), I'm sure it wouldn't
be too difficult to convince the Savane authors to slap the same label
on Savane, since it's already pretty much mostly used by GNU.

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Stefan Monnier]

> If Bzr had required a heavy investment of development before it would be
> meet the technical needs of the Emacs project, then the decision would
> have been different.

Bzr did need a significant investment (which lasted more than a year)
before it would (barely) meet the technical needs of the Emacs project.

But in any case, this discussion has nothing to do with Emacs:
this decision was not taken by the Emacs maintainers but by Richard.
Please move it elsewhere,


        Stefan

Re: DVCSes, bug trackers, and GNUness

[Stefan Monnier]

> No, let's.  Because I know you're tired of it, and I know why you are:
> It's the decision everyone hated but taken only because of GNUness.

No, I'm tired of it because it's just bikeshedding.

As for the politics behind it, I'm not interested in going there.


        Stefan

Re: DVCSes, bug trackers, and GNUness

[Karl Fogel]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
>> If Bzr had required a heavy investment of development before it would be
>> meet the technical needs of the Emacs project, then the decision would
>> have been different.
>
>Bzr did need a significant investment (which lasted more than a year)
>before it would (barely) meet the technical needs of the Emacs project.

Sorry; I forget the actual complexity of the transition.  We did not
actually do the switch before Bzr was ready ("barely", perhaps), and
furthermore the investment resources to get it to where we needed it
were more or less known to be committed to that purpose, when we made
the decision.  I guess Jordi is proposing the same for Savane, though.

>But in any case, this discussion has nothing to do with Emacs:
>this decision was not taken by the Emacs maintainers but by Richard.

The second statement is true, but I'd disagree with the first :-).
On the other hand, like you said, http://blue.bikeshed.com/.

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

On 25 April 2012 16:28, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
> Bzr did need a significant investment (which lasted more than a year)
> before it would (barely) meet the technical needs of the Emacs project.

Right. I want the same for Savane.

> But in any case, this discussion has nothing to do with Emacs:
> this decision was not taken by the Emacs maintainers but by Richard.

Eh? The decision for Emacs  to use bzr has got nothing to do with
Emacs. Are you saying Richard has got nothing to do with Emacs?

> Please move it elsewhere,

Where to?

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Alan Mackenzie]

On Wed, Apr 25, 2012 at 04:58:37PM -0400, Jordi Gutiérrez Hermoso wrote:
> On 25 April 2012 16:28, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
> > Bzr did need a significant investment (which lasted more than a year)
> > before it would (barely) meet the technical needs of the Emacs project.

> Right. I want the same for Savane.

> > But in any case, this discussion has nothing to do with Emacs:
> > this decision was not taken by the Emacs maintainers but by Richard.

> Eh? The decision for Emacs  to use bzr has got nothing to do with
> Emacs. Are you saying Richard has got nothing to do with Emacs?

> > Please move it elsewhere,

> Where to?

gnu-misc-discuss@gnu.org would seem taylor made for this discussion.

> - Jordi G. H.

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: DVCSes, bug trackers, and GNUness

[Richard Stallman]

Savannah and Bzr are two practical issues with very little in common.
You seem to be proposing a very simplistic uniform way of approaching
all issues, but that's a misguided goal.

    > Savannah isn't a package, it's a facility.  So it is a different kind
    > of issue.

    I don't understand the distinction. Can you explain?

The reason I decided Emacs should use Bzr is to encourage development
of Bzr, the software package.  Savanna is not a software package.
It is a repository service.

Savannah is run by overworked volunteers, and each package means a
little more work for them.  Thus, the number of packages they can host
is limited.  So even if there were a motive to ask people "please,
everyone, use Savannah, not other repository services", it would not
be feasible to do it.

Users are not a limiting factor for Savannah development.  Savannah
has plenty of users.  So there is no need for us to "send some users
to Savannah".

    I do, but I would like to coerce other people to help me, just like
    people are being coerced right now to work with bzr

Nobody is being coerced to work on Bzr development, and nobody will be
coerced to work on Savannah development.  But we can try to recruit
more people to work on Savannah development.

--
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call

Re: DVCSes, bug trackers, and GNUness

[Les Harris]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:

> I do, but I would like to coerce other people to help me, just like
> people are being coerced right now to work with bzr and hopefully
> improve it.

I hadn't realized savannah needed more help.  Interestingly, the way
Savannah describes itself would seem to agree with what you are saying
in general:

> Savannah aims to be a central point for development, maintenance and
> distribution of official GNU software.

What does one have to do to get involved?  Is the information at
http://savannah.gnu.org/maintenance/HowToBecomeASavannahHacker still
current?

As a FSF member and one who has contributed to free software projects I
certainly wouldn't mind donating hours to a worthy cause.

-- 
Do they only stand
By ignorance, is that their happy state,
The proof of their obedience and their faith?

Re: DVCSes, bug trackers, and GNUness

[Richard Stallman]

    >  A: A whole lot nicer than it was four years ago

    That was in fact a secondary reason to choose it, IIRC. To push it to improve.

That's exactly why I considered the issue imprtant and insisted:
to make sure that Bzr would improve.


--
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call

Re: DVCSes, bug trackers, and GNUness

[Richard Stallman]

    Okay, so I'm getting a consistent message that if I want people to
    prefer Savane over debbugs, I need to first make sure Savane can have
    an email and SOAP interface. Is this accurate?

That may be true.

						   Then it's a GNU project
    with comparable features to debbugs,

The term "a GNU project" seems to be causing misunderstanding.  Emacs
and Bzr are GNU packages, but Savannah is not a GNU package.  It is
not a software package at all.

Savane is a software package, but it isn't a GNU package.  We use it,
and I suppose the Savannah hackers help maintain it, but I don't know
the details of how that is done or how they relate to other developers
of Savane.

Glenn reminded us about GNATS.  GNATS is a GNU package that was
developed in the early 90s.  I don't know how it relates to modern-day
needs.  I will ask its maintainers how it stands.


--
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call

Re: DVCSes, bug trackers, and GNUness

[Richard Stallman]

If the Savane developers would like to make it a GNU package,
I invite them to talk with me about it.


--
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call

Re: DVCSes, bug trackers, and GNUness

[Richard Stallman]

gnu-misc-discuss is a place where people who disagree with the goals
of GNU argue about them.  We should not use it for discussions of what
to do in the GNU Project.

I suggest that people interested in improving Savannah discuss it
on a Savannah list.

--
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! That's nonfree (freedom-denying) software.
  Use Ekiga or an ordinary phone call

gnu-misc-discuss [ was: DVCSes, bug trackers, and GNUness]

[Alan Mackenzie]

Hello, Richard.

On Thu, Apr 26, 2012 at 08:09:09AM -0400, Richard Stallman wrote:
> gnu-misc-discuss is a place where people who disagree with the goals
> of GNU argue about them.  We should not use it for discussions of what
> to do in the GNU Project.

It used to be such a pit, but the two trolls, one anonymous, (and a few
hangers on) who smothered the list in foul mouthed abuse appear to have
given up.  I think they were being paid by proprietary companies to
damage the ethos of free software as much as possible.

Now, there is only one regular poster, somebody spamming meetings of a
free software society in New York, and he's been doing this for many
years.

Otherwise the list is as quiet as a morgue.

> --
> Dr Richard Stallman

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: DVCSes, bug trackers, and GNUness

[Glenn Morris]

Stefan Monnier wrote:

> [...] the main weaknesses AFAICT of Debbugs being its limited
> search&classification functionality, as well as [...]

BTW, do you know about
http://debbugs.gnu.org/cgi/search.cgi ?

It's not perfect, but I can generally find what I want to find.

I don't know what you mean by "classification". Maybe explain on
help-debbugs@gnu if you think there is a chance of something being
implemented to make it better. Small tweaks are doable; radical changes
are of course unlikely.

Re: DVCSes, bug trackers, and GNUness

[Steinar Bang]

>>>>> Juanma Barranquero <lekktu@gmail.com>:

> I define "git fanboys" as people in one of these two categories: a)
> those that would consider anything by Linus Torvalds as a gift from
> heaven;

FWIW Linus' snotty attitude towards existing VCSes kept me away from git
for a long time.

> b) those whose argument (really, they only have one) to prefer git is
> that it is "fast".

Hmno... my main argument for using git is the way it does branching and
merging. 

In fact one reason _not_ to use it is the opposite of "fast". That
reason is that it is so dog slow on Windows.  My old debian PII 233MHz
home network "server", outperforms my 64bit, 8 core monster at work on
git operations (the operations I regularily use, ie. git-fetch,
git-status, git-gc, git-add, git-commit, git-push).

> People with a more nuanced vision of dVCS are excluded from my
> definition and need not feel offended by it.

People keep lumping dVCSes together, but the distinguishing feature for
git isn't that it is a dVCS, but the way it does branching and merging.
It's really quite, quite, clever. 

(See http://progit.org/book/ch3-0.html for more information)

Re: DVCSes, bug trackers, and GNUness

[Jordi Gutiérrez Hermoso]

On 29 April 2012 13:07, Steinar Bang <sb@dod.no> wrote:
> People keep lumping dVCSes together, but the distinguishing feature for
> git isn't that it is a dVCS, but the way it does branching and merging.
> It's really quite, quite, clever.

No, git does nothing special in this regard that other DVCSes don't
already do. Have you tried them?

- Jordi G. H.

Re: DVCSes, bug trackers, and GNUness

[Miles Bader]

Jordi Gutiérrez Hermoso <jordigh@octave.org> writes:
>> People keep lumping dVCSes together, but the distinguishing feature
>> for git isn't that it is a dVCS, but the way it does branching and
>> merging.  It's really quite, quite, clever.
>
> No, git does nothing special in this regard that other DVCSes don't
> already do. Have you tried them?

Please give it a rest.  This issue does nothing but generate flames on
this list.

-miles

-- 
"Suppose He doesn't give a shit?  Suppose there is a God but He
just doesn't give a shit?"  [George Carlin]