This is the mail archive of the cygwin mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: zsh-4.1.1-2 still seems broken under cygwin-1.5.7-1.


On Thu, 12 Feb 2004, Larry Hall wrote:

I have to preface this by saying it's quite long and very OT.  If you
have something better to do, like fixing bugs, by all means skip reading
the rest of this.

> At 01:21 PM 2/12/2004, Peter A. Castro you wrote:
> >On Thu, 12 Feb 2004, Christopher Faylor wrote:
> >
> >> On Thu, Feb 12, 2004 at 09:44:26AM -0500, Larry Hall wrote:
> >> >As I mentioned before, it's better to verify that the current snapshot does
> >> >address the problem you're seeing locally.  Otherwise, if you're seeing a
> >> >variant or something different than the rest, your problem won't be known
> >> >until after 1.5.8 is released.  See <http://cygwin.com/snapshots/> for the
> >> >latest snapshots you can download and try.
> >>
> >> Right.  This is *exactly* why there are problems with zsh.  Someone had
> >> an easily reproducible problem that showed up in 1.5.6.  Rather than
> >> report it they apparently waited for 1.5.7 to show up, assuming that all
> >> of their problems would be solved.
> >
> >Well, I would have reported it, but I never had a chance to upgrade to
> >1.5.6 in the first place (have been too busy with other things), and then
> >along came 1.5.7 any all hell broke loose.  For anyone who wants a stable
> >environment, what's the harm in stepping back a version or two so that
> >they can get back to work?
>
> I think I answered this in my prior response but I'll clarify.  I'm not
> recommending that people jump head-first into using the latest snapshot
> and never look back no matter how much trouble they find.  It's prudent and
> recommended to have a backup that you're comfortable with.  If
> you find a problem with the snapshot that significantly impacts your
> productivity, report it, roll back, and do what you need to do.  It's not
> an all or nothing affair.  Of course, if you can help debug, that's great
> too!  Sounds like you recognize all this but I'm just a little concerned

Recognize?  Hell, I *do* debug!  But, debugging isn't for everyone and
everyone shouldn't be encouraged to attempt debugging unless they have
the skills to do so.

> that you're giving the impression to others that once you step into the
> realm of a snapshot, you can't roll back.

I'm not trying to debate you on this issue.  I feel snapshots are a good
thing to try, *if you can afford to do it*.  I never stated that you
couldn't rollback from a snapshot, given one saves the original version
first (and even that's not a requirement), but doing all that is extra
effort in the eyes of a non-dev user.  By contrast, there have been many
snapshots which break in various ways in which case rolling back is the
only feasible recourse.

I was thinking more along the lines of someone who's been given the
cygwin environment to work in, by someone else, and who tried an updated
(why? because the could or someone else told them to do so!), but found
problems and really doesn't know what to do next.  For that person,
simply rolling back to a stable version is more expedient and less hassle
than hunting for a snapshot and trying it out.  Not everyone wants to be
a guinea pig, nor should everyone have to be one.  This is how my work
environment is, btw.  We have several developers who are simply "cygwin
users" and don't know the details of how/why cygwin is, but simply need
it to get their jobs done.  I am a resource for them, but I'm pretty busy
myself, and rather than guide them through trying a snapshot or having
them slog through email archives or get cygwin email themselves, it's
easier to tell them "rollback" so they can continue their work.  Once
I've seen that their problems are addressed (usually by my own testing,
hey, I try and keep track of these things), I recommend they upgrade when
they have a moment.

> >I know that when things like this break in my
> >*work* environment, the last thing I'd want is to play russian roulette
> >with a snapshot which might make things worse.
>
> And if it does, so what?  Report, roll back, move on.  What's so hard
> about that?

Nothing, but if you're just trying to get back to "working" and aren't an
active developer/maintainer, and don't have the time to test, then why
should you try an untested snapshot?  I know plenty of people who, on the
advice of others, try various pre-releases, invariably trash their setup
and have to re-install to fix it.  And this isn't just cygwin, but applys
to cygwin as well.  Unless I know the person in question and what they
are capable of, I tent to direct them to the safest option first.

> >If you're not a
> >developer, why not just wait?
>
> You certainly can but then you run the risk of your problem not getting
> fixed even if others think the problem has been fixed.  Actually, I would
> characterize that stance as russian roulette but if that's what you want
> to do and it works for you, I can't stop you.

Why would you want to stop me in the first place?  Do you think you know
what's best for me?  How about for everyone?  How about the guy who's on
deadline to get a product out and desperately needs his cygwin
environment to work?  No, russian roulette is the act of willfully trying
something else and accepting the consequences, (even if fatal :).  If I
stay on a specific release that's called "playing it safe".  Witness all
the people who are still on "B20".  Most people are of a static-nature.
If it isn't broke and it works for what you need, why change?  Not all
change is good.

> >For the life of me, I could never quite
> >understand why you and Larry keep pushing the latest stuff when it hasn't
> >been fully tested or released.
>
> And how do you expect it to get tested and ready for release without
> people's help?  That's my point (and I think Chris would concur).

By those people who are interested in testing, of course, and you're
ignoring my point.  I did test it, but my note was lost in the ether (see
other email on this topic).  For the vast majority of users, all they
want is something that works with a minimum of hassle, where
  hassle < "finding latest snapshot, downloading compressed dll,
uncompressing it, copying old dll to a safe place and copying new dll in
place of original, testing, finding it breaks, copying old dll back".

Can you guarantee a given snapshot is bug-free?  Didn't think so, and
neither can I.  Just because a snapshot might work for me, doesn't mean
it'll work for someone else.  We've all witnessed the phenomena of "works
for me, but not my friend".  Sometimes the risk just isn't worth it.

> >Now, that being said, I am a developer
> >and I do play with the latest snapshots from time to time (like in this
> >case), but that's in my home environment where I can tollerate things
> >breaking (and often break them myself for fun :).
>
> Everyone has to figure out where and how it makes sense to work with new
> software (and not just Cygwin).  If running with newer stuff at home while
> keeping older stuff at work is what works for you, I won't argue.  But it
> does seem like you're pushing too hard on this notion that one cannot recover
> from a snapshot that doesn't work for them.  This is clearly false.

I'm not pushing this notion at all!  Where, in all of this, am I actually
pushing or even implying that you can't recover from a broken snapshot?
Show me!  Please see earlier comments concerning getting work done with a
minimum of hassle and in a timely fashion.  Playing the "Upgrade Game" is
costly in many, many ways, however.  Do you update *all* of your software
on every machine to the latest snapshot all the time?  Neither does
anyone else, so why are you asking others to do so?  Why do you have a
problem with people waiting for official releases to upgrade?  Is it
because those people aren't actively contributing to the betterment of
cygwin?  Not everyone is you or I or, especially, cgf.  I hate to tell
you, but lots of people are only capable of using what's available
without contributing back.  There are a whole slew of social factors to
considder in this and it's not clear cut as black and white, and usually
has to do with time and resources and effort.

> >> There are still problems with the latest snapshot that I hope to have
> >> fixed today.  We'll see.
> >
> >And, that little statement would make me even more nervious about getting
> >back to something stable if I had a work schedule to keep.
>
> That's OK.  There are still known problems, as the email archives can
> attest.  If those aren't an issue for you, then running with the latest
> snapshot is a good idea.  If not, then wait for the next one.  Snapshots
> aren't just generated willy-nilly.  They're only generated when someone,

I don't know... I've seen cgf generate several snapshots back-to-back
almost, so it appears, without thorough testing, so in some
circumstances, yes they are generated willy-nilly.  But I certainly
understand the process and it's not the process I'm questioning.  It's
whats good for the users that I question.

> generally Chris, thinks it's worthwhile to have something to test by
> the community at large.  It's an attempt to get releases that are very

"community at large" != "everyone"

Get it?  One size does not fit all.  I get private email from people who
aren't part of the cygwin list, saying they have this hanging problem and
asking me what to do.  I don't know what they are capable of, I don't
have the time to outline something detailed and I hate to just point
people at websites because they are usually after a specific thing and
will have to invest a chunk of their time to find out all about snapshots
and how to use them.  I prefer to give people as much help as thoroughly
as possible.  If the short answers gets them going faster, great.  If
they don't like the answer, or are curious to know more, then I always
advise them to do the research necessary, but my goal is to get them
going as quickly as possible.

> stable and generally useful.  If no one uses them, then releases suffer.
> But no one can complain about the low quality release then. ;-)

I'm not complaining (Ok, perhaps I am in a way).  I've always felt the
quality of work is quite good in cygwin, (and that goes for even the
things that break), and that's not what I have an issue with.  If noone
uses the release, then perhaps that's an indication that the software has
reached the end of it's usefulness and doesn't deserve being tested.
Fortunately that's not the case with cygwin, or we wouldn't be having
this spirited debate right now.  I tend to error on the side of caution
because I've seen too many cases of people getting crushed by the
upgraded boulders.  If that's unsatisfactory to you, then I'm sorry, but
I'm just trying to help people the best I can.

Now, this is severly OT.  Please make one more reply, if you wish, and
let the thread die.  Aren't there some bugs we all need to be working on?
:)

-- 
Peter A. Castro <doctor@fruitbat.org> or <Peter.Castro@oracle.com>
	"Cats are just autistic Dogs" -- Dr. Tony Attwood

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]