View mode: basic / threaded / horizontal-split · Log in · Help
June 14, 2008
Re: D learning curve
Jarrett Billingsley wrote:
> It *has* a standard library.  It's Phobos.  It just so happens that 
> many D users use Tango instead ;)  Of course that begs the question: 
> is a "standard library" that is used by a minority all that 
> "standard"?
                                            ^^^^^^^^
Do you have any proof other than the very unrepresentative poll in this
newsgroup that Tango has more users than Phobos? As far as I know no one
knows the ratio of Phobos and Tango users, so this statement is only an
assumption, based on what exactly? The hopes of Tango followers?

I don't mean to affront you or any other Tango user, but I respond very
harsh if I'm under the impression that someone uses propaganda to push
his goal.

Are there any download counts published that could give us a *hint* of
which library is used more often?

LLAP,
Sascha
June 14, 2008
Re: D learning curve
"Jarrett Billingsley" <kb3ctd2@yahoo.com> wrote in message 
news:g2unqe$18lg$1@digitalmars.com...
>
> It *has* a standard library.  It's Phobos.  It just so happens that many D 
> users use Tango instead ;)  Of course that begs the question: is a 
> "standard library" that is used by a minority all that "standard"?

I might be off-base, but I've always thought of it like this:

Phobos: Official Standard
Tango: De Facto Standard

Of course, Phobos being the only one that supports D2 right now kinda 
changes things for the moment.
June 14, 2008
Re: D learning curve
Sascha Katzner wrote:

> Jarrett Billingsley wrote:
>> It *has* a standard library.  It's Phobos.  It just so happens that
>> many D users use Tango instead ;)  Of course that begs the question:
>> is a "standard library" that is used by a minority all that
>> "standard"?
>                                              ^^^^^^^^
> Do you have any proof other than the very unrepresentative poll in this
> newsgroup that Tango has more users than Phobos? As far as I know no one
> knows the ratio of Phobos and Tango users, so this statement is only an
> assumption, based on what exactly? The hopes of Tango followers?
> 
> I don't mean to affront you or any other Tango user, but I respond very
> harsh if I'm under the impression that someone uses propaganda to push
> his goal.
> 
> Are there any download counts published that could give us a *hint* of
> which library is used more often?
> 
> LLAP,
> Sascha

You could look at the library code and projects at dsource that is being
written. Most projects and almost every big project uses Tango. That is
maybe even more important than the number of users.
June 14, 2008
Re: D learning curve
"Sascha Katzner" <sorry.no@spam.invalid> wrote in message 
news:g2vrqi$ui1$1@digitalmars.com...
> Jarrett Billingsley wrote:
>> It *has* a standard library.  It's Phobos.  It just so happens that many 
>> D users use Tango instead ;)  Of course that begs the question: is a 
>> "standard library" that is used by a minority all that "standard"?
>                                             ^^^^^^^^
> Do you have any proof other than the very unrepresentative poll in this
> newsgroup that Tango has more users than Phobos? As far as I know no one
> knows the ratio of Phobos and Tango users, so this statement is only an
> assumption, based on what exactly? The hopes of Tango followers?
>
> I don't mean to affront you or any other Tango user, but I respond very
> harsh if I'm under the impression that someone uses propaganda to push
> his goal.
>
> Are there any download counts published that could give us a *hint* of
> which library is used more often?
>
> LLAP,
> Sascha

I'm mostly going off of what Lutger has said, as well as a somewhat 
subjective rating based on my experience on the D IRC channel.

As for download counts, I'm not so sure any real comparison could be made. 
Phobos comes with the compiler, which can be obtained from many sources 
(directly from DM, the GDC sourceforge page, debian packages).  Also, Tango 
can be checked out of SVN anonymously so I don't think there's any reliable 
way of knowing how many people have it.
June 14, 2008
Re: D learning curve
Lutger wrote:
> You could look at the library code and projects at dsource that is
> being written. Most projects and almost every big project uses Tango.
> That is maybe even more important than the number of users.

If a specific project is significant or not is a highly subjective 
decision and the pure count of projects which depend on either Phobos or 
Tango doesn't say much about the real spread or significance of either one.

If you want to use other libraries and projects as a benchmark, you have 
to weight them according to an objective criteria, such as their users 
and their relevance... which seems to me more complex than the original 
question.

And even if you successfully master this you don't account for a lot of 
people who don't use any third party libs.

LLAP,
Sascha
June 14, 2008
Re: D learning curve
Jarrett Billingsley wrote:
> As for download counts, I'm not so sure any real comparison could be
> made. Phobos comes with the compiler, which can be obtained from many
> sources (directly from DM, the GDC sourceforge page, debian
> packages).  Also, Tango can be checked out of SVN anonymously so I
> don't think there's any reliable way of knowing how many people have
> it.

I think it is pretty safe to assume that the majority of Tango users
prefer the pre-bundled package and not the subversion way (the same
should be true for DMD/Phobos). And since the compiler is already
included in this release you don't have to subtract this count from the
one of the DM website.

So I think if you account for the same time period the download counts
should be a very good hint for the usage ratio of the two libraries.

LLAP,
Sascha
June 14, 2008
Re: D learning curve
On Sat, 14 Jun 2008 12:18:00 +0200, Lutger wrote:

> Sascha Katzner wrote:
> 
>> Jarrett Billingsley wrote:
>>>..You could look at the library code and projects at dsource that is 
being
> written. Most projects and almost every big project uses Tango. That is
> maybe even more important than the number of users.

Which is[was?] a real problem as, for me at least, tango keeps breaking 
with each new release of D.  And I didn't find DSSS to be all that 
workable either.  DMD works fine, and so does Phobos.  Tango was unending 
problems.  And I frequently switch to DMD2 for a new release (and 
sometimes switch back to DMD), so Tango isn't even consistently an option.

Perhaps some of the problems of which I'm complaining have been fixed.  I 
last checked over 6 months ago.  But I'm not real inspired to try it out 
again, either.  If I wanted to spend all my time fighting with my 
computer I'd install Gentoo.

The upshot is that if a project requires Tango, I generally assume that 
if I try to use it I'll end up spending all my time in compilation and 
configuration, and figuring out why what I tried didn't work.  I don't 
know what configurations the Tango people expect a system to have, but 
mine doesn't have them.  Once I tried setting up a special user who only 
executed DMD1.x (forget which version) with Tango.  After 3-4 days I gave 
that up as a bad job.  I didn't even know why it wasn't working.

OTOH, I've got to admit that many people seem to really like Tango.  And 
I have no clue as to what the differences between our systems are.  
(Though there's probably typically so many differences that even that 
wouldn't help much.)
June 15, 2008
Re: D learning curve
Charles Hixson wrote:
> On Sat, 14 Jun 2008 12:18:00 +0200, Lutger wrote:
> 
>> Sascha Katzner wrote:
>>
>>> Jarrett Billingsley wrote:
>>>> ..You could look at the library code and projects at dsource that is 
> being
>> written. Most projects and almost every big project uses Tango. That is
>> maybe even more important than the number of users.
> 
> Which is[was?] a real problem as, for me at least, tango keeps breaking 
> with each new release of D.  And I didn't find DSSS to be all that 
> workable either.  DMD works fine, and so does Phobos.  Tango was unending 
> problems.  And I frequently switch to DMD2 for a new release (and 
> sometimes switch back to DMD), so Tango isn't even consistently an option.
> 
> Perhaps some of the problems of which I'm complaining have been fixed.  I 
> last checked over 6 months ago.  But I'm not real inspired to try it out 
> again, either.  If I wanted to spend all my time fighting with my 
> computer I'd install Gentoo.
> 
> The upshot is that if a project requires Tango, I generally assume that 
> if I try to use it I'll end up spending all my time in compilation and 
> configuration, and figuring out why what I tried didn't work.  I don't 
> know what configurations the Tango people expect a system to have, but 
> mine doesn't have them.  Once I tried setting up a special user who only 
> executed DMD1.x (forget which version) with Tango.  After 3-4 days I gave 
> that up as a bad job.  I didn't even know why it wasn't working.

I've never had problems this bad, but I am disappointed that Tango 
releases make no attempt to be backwards compatible (they deprecate some 
syntax for one release and then get rid of it). Already, the simplest 
code in the book is not working. Pretty soon, anything that depends on 
the old collection packages is going to break. If you want to use two 
libraries that depend on different versions of Tango, you're going to be 
spending some time fixing them up.

Of course, Phobos 2 might have this problem, I'm not sure.

> OTOH, I've got to admit that many people seem to really like Tango.  And 
> I have no clue as to what the differences between our systems are.  
> (Though there's probably typically so many differences that even that 
> wouldn't help much.)

The GC/runtime/IO is faster & less buggy than Phobos's in general, so 
for production code (or code you want to be blazing fast), Tango is th 
better choice. The upper-level library also has a MUCH nicer API, IMO 
(you can use it Java-style, like I do, or C/free-function style if you 
prefer, while Phobos is a hodgepodge of different APIs that don't fit 
together as well).
June 15, 2008
Re: D learning curve
Charles Hixson wrote:
> On Sat, 14 Jun 2008 12:18:00 +0200, Lutger wrote:
> 
>> Sascha Katzner wrote:
>>
>>> Jarrett Billingsley wrote:
>>>> ..You could look at the library code and projects at dsource that is 
> being
>> written. Most projects and almost every big project uses Tango. That is
>> maybe even more important than the number of users.
> 
> Which is[was?] a real problem as, for me at least, tango keeps breaking 
> with each new release of D.  And I didn't find DSSS to be all that 
> workable either.  DMD works fine, and so does Phobos.  Tango was unending 
> problems.  And I frequently switch to DMD2 for a new release (and 
> sometimes switch back to DMD), so Tango isn't even consistently an option.
> 
> Perhaps some of the problems of which I'm complaining have been fixed.  I 
> last checked over 6 months ago.  But I'm not real inspired to try it out 
> again, either.  If I wanted to spend all my time fighting with my 
> computer I'd install Gentoo.
> 
> The upshot is that if a project requires Tango, I generally assume that 
> if I try to use it I'll end up spending all my time in compilation and 
> configuration, and figuring out why what I tried didn't work.  I don't 
> know what configurations the Tango people expect a system to have, but 
> mine doesn't have them.  Once I tried setting up a special user who only 
> executed DMD1.x (forget which version) with Tango.  After 3-4 days I gave 
> that up as a bad job.  I didn't even know why it wasn't working.
> 
> OTOH, I've got to admit that many people seem to really like Tango.  And 
> I have no clue as to what the differences between our systems are.  
> (Though there's probably typically so many differences that even that 
> wouldn't help much.)

What I do is pick a version of the compiler that works and stick with 
that.  As long as you don't need newer versions of tango or other apis 
that works fine for mw.  I sometimes try out updates however if they 
fail, or are too tricky to fix, I rollback.

-Joel
June 15, 2008
Re: D learning curve
Robert Fraser wrote:
> The GC/runtime/IO is faster & less buggy than Phobos's in general, so 
> for production code (or code you want to be blazing fast), Tango is th 
> better choice.

If you want "blazing fast" code, it is mostly the better choice to use 
customized code. Libraries tend to be over protective (for a good 
reason) and therefore sometimes wasting a lot of time with unnecessary 
stuff like error checking for errors that could never occur in a 
specific situation. Tasks like memory management and IO are very good 
examples for this.

When I'm using a library I don't expect to get the fastest code, but 
_reliable_ code that I don't have to write on my own. Your observation 
that Phobos seems to have more bugs _could_be_ a direct consequence of 
that it is used by more people... SCNR ;)

LLAP,
Sascha
1 2 3
Top | Discussion index | About this forum | D home