>>>>> Simon Michael <[email protected]> writes:
> First, an apology. John, I'm sorry for being an annoyance on #ledger last
> friday. As a "house guest" on Ledger's IRC channel, I always try to watch
> what I say. I picked a bad time to mention hledger, and to trigger a
> conversation about relative bugginess. Sorry about that.
I didn't think you were being an annoyance Simon! I've always enjoyed your
presence in the channel and am a bit saddened that I may see less of you, now
that you're moving on to #hledger... But that's life.
> Next I learned that opinions of mine such as
> 1. Ledger users continue to experience a lot of bugs[1], and
> 2. hledger is successfully providing a less buggy (less featureful, less
> performant) experience are not generally accepted as the current state of
> things (gasp!). In fact they are controversial.
Of course there are always bugs in software, and C++ makes it especially hard
to crush them once and for all. This is why I rely on the Boost libraries so
much -- to try to win back some of the benefits which are automatic in a
language like Haskell.
That being said, what constitutes a bug, how buggy software is overall, and
how buggy one perceives that software to be, are all very subjective, even
when you throw numbers into the mix. The plain fact is that we can't know
everything objectively, and no user is ever going to use every possible
capability of the system. I probably use less than 50% of what ledger is
capable of myself on a regular basis.
As an example, I once had an argument on IRC about the relative stability of
ZEVO's ZFS implementation on Mac, and the zfsonlinux project. My experience,
born out for over a year, was simple:
1. I'd never experienced a single crash or loss of data with ZEVO.
2. I was seeing kernel panics at least once a week using zfsonlinux, and
there were certain commands (rsync) which would not only lock up my
filesystem, but make it unmountable after a reset. I lost gigabytes of
data this way.
Easy argument to make, right? It turns out, not at all. My experience was
not everyone's experience. Despite the fact that ZEVO was bullet-proof for
me, there was a whole host of people having problems with it and thanking
their stars for the stability that moving to zfsonlinux gave them. I argued
my point, but eventually lost, being called "anecdotal" and "an exceptional
case". Who am I to say that I wasn't?
It turns out I was doing funny stuff with extended attributes, and was using a
ZEVO partition I had cloned on the zfsonlinux machine -- all use cases which
very few others had even attempted. These bugs were later fixed, but they
created a perception which changed the way I evaluated both packages for a
long time.
My point is that "bugginess" is a dangerous term to sling around. I would
consider hledger's use of floating-point[1] a more serious bug than every one
currently in Ledger's bug tracker. And yet other users may go for years and
never have a problem with it.
So I think it makes no sense to debate such things, or even to compare them.
Rather we should keep collecting our bug reports, assigning priorities, and go
about combating entropy as best we're able.
Thanks for everything you've contributed to the Ledger-verse,
John
[1] In fact, I would like to work with hledger to contribute code to solve
this. I care more about the Ledger concept prevailing over other, worse
alternatives, than I do about what language the most-used version is
written in.
--
---
You received this message because you are subscribed to the Google Groups
"Ledger" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
