Bug#840908: Uscan's Sourceforge reflector is too naive

2016-10-16 Thread Adrian Bunk
On Sat, Oct 15, 2016 at 11:06:50PM -0500, Steve M. Robbins wrote:
> On Sunday, October 16, 2016 11:48:32 AM CDT Paul Wise wrote:
>...
> > Your upstream isn't naming snapshot tarballs correctly. This should be
> > fixed either in boost upstream 
> 
> I know this is the popular Debian perception and certainly it is a nuisance 
> that the filenames are not unique.  All I will say is that the folks 
> releasing 
> Boost are not novices and likely have a defensible reason for their madness.  
>...

This is not a Debian perceptions, it is just bad when published tarballs 
with different names carry the same contents.

And based on real-life experience in companies, I can tell you that even 
"we are doing it this way for 20 years" does not necessarily imply that 
there is (or ever was) any defensible reason...

The uscan issue might be fixed due to quick action by Paul, but could 
you anyway ask upstream to switch to a unique naming scheme?

It is really hard to imagine any defensible reason for naming the daily 
snapshot from master boost_1_62_0.tar.bz2 instead of something like 
boost_1_62_0_master_20161015.tar.bz2

> Best,
> -Steve

cu
Adrian

-- 

   "Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
   "Only a promise," Lao Er said.
   Pearl S. Buck - Dragon Seed



Bug#840908: Uscan's Sourceforge reflector is too naive

2016-10-15 Thread Paul Wise
On Sun, Oct 16, 2016 at 11:48 AM, Paul Wise wrote:

> This should be ... or in the boost watch files. Modifying
> the boost watch file is dependent on the fix for the above issue.

I've now committed the fix, you can use something like this:

version=3
opts="uversionmangle=s/_/./g,dversionmangle=s/\+dfsg$//" \
http://sf.net/boost/ .*/[.\d]+/boost_([\d_]*)\.tar.bz2

Note the space after the redirector URL.
Note the numerical directory match, it needs adjusting if you want alpha/beta.

-- 
bye,
pabs

https://wiki.debian.org/PaulWise



Bug#840908: Uscan's Sourceforge reflector is too naive

2016-10-15 Thread Steve M. Robbins
On Sunday, October 16, 2016 11:48:32 AM CDT Paul Wise wrote:
> On Sun, Oct 16, 2016 at 10:24 AM, Steve M. Robbins wrote:
> > My suggestion is that the ones with "snapshots" in the path are simply
> > filtered out from list displayed by the reflector as these are not
> > release files.
> 
> That sounds like an ugly hack that I would rather not see implemented.

I can't disagree.  :-)  
If I knew how to fix it in the watch file, I would do so.


> The are two issues here:
> 
> The redirector was designed in the days when there were no directories
> on the sf.net files section.

I see.  One other thing I will mention is that the current reflector does
show two occurrences of "boost_1_62_0.tar.bz2".  (And just now it even shows 
the path and a "direct link" that I'm pretty sure weren't present earlier 
today!)  However, both occurrences link to the same reflector URL [1] that ends 
up at the wrong URL on sf.net.

[1] https://qa.debian.org/watch/sf.php/boost/boost_1_62_0.tar.bz2


> Your upstream isn't naming snapshot tarballs correctly. This should be
> fixed either in boost upstream 

I know this is the popular Debian perception and certainly it is a nuisance 
that the filenames are not unique.  All I will say is that the folks releasing 
Boost are not novices and likely have a defensible reason for their madness.  
More to the point: it is out there and if uscan can be made more flexible, it 
would be appreciated.

> or in the boost watch files. Modifying
> the boost watch file is dependent on the fix for the above issue.

OK.  Thanks for looking into this!

Best,
-Steve


signature.asc
Description: This is a digitally signed message part.


Bug#840908: Uscan's Sourceforge reflector is too naive

2016-10-15 Thread Paul Wise
On Sun, Oct 16, 2016 at 10:24 AM, Steve M. Robbins wrote:

> My suggestion is that the ones with "snapshots" in the path are simply
> filtered out from list displayed by the reflector as these are not
> release files.

That sounds like an ugly hack that I would rather not see implemented.

The are two issues here:

The redirector was designed in the days when there were no directories
on the sf.net files section. I'm going to try adding links that
include the path, existing links will need to remain for backwards
compatibility.

Your upstream isn't naming snapshot tarballs correctly. This should be
fixed either in boost upstream or in the boost watch files. Modifying
the boost watch file is dependent on the fix for the above issue.

-- 
bye,
pabs

https://wiki.debian.org/PaulWise



Bug#840908: Uscan's Sourceforge reflector is too naive

2016-10-15 Thread Steve M. Robbins
Package: qa.debian.org
Severity: normal

The uscan download from sourceforge doesn't download what you expect
for boost.  The reason is that the link provided by the reflector page
[1] is incorrect: it leads to a "snapshot" url [2].  The correct
URL is [3].

Paul Wise indicated [4] that the reflector is a proxy for the RSS
feed and indeed both URLs are present:

  
https://sourceforge.net/projects/boost/files/boost/snapshots/master/boost_1_62_0.tar.bz2/download
  
https://sourceforge.net/projects/boost/files/boost/1.62.0/boost_1_62_0.tar.bz2/download

My suggestion is that the ones with "snapshots" in the path are simply
filtered out from list displayed by the reflector as these are not
release files.


[1] https://qa.debian.org/watch/sf.php/boost/
[2] 
https://sourceforge.net/projects/boost/files/boost/snapshots/master/boost_1_62_0.tar.bz2/download
[3] 
https://sourceforge.net/projects/boost/files/boost/1.62.0/boost_1_62_0.tar.bz2/download
[4] https://lists.debian.org/debian-devel/2016/10/msg00292.html


-- System Information:
Debian Release: stretch/sid
  APT prefers unstable
  APT policy: (500, 'unstable')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 4.7.0-1-amd64 (SMP w/4 CPU cores)
Locale: LANG=en_CA.UTF-8, LC_CTYPE=en_CA.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)