Open-source ...typical

Miscellaneous Topics
#1

OpenOffice has just done a point release - 1.1.0 to 1.1.1 - which fixes a lot of major bugs (mainly to do with failing to render documents and inability to load or save file formats correctly). I was looking forward to using it, because several of the fixed bugs have - literally - prevented us from editing particular word documents.

Only, in this allegedly minor point release, they have:

  • made the 1.1.0 install procedure no longer work; if you try what you did with 1.1.0, then it appears to install but you end up with non-working crud that just crashes with a completely inexplicable random error message and no explanation.
  • included an auto-detect prior install that … doesn’t. And doesn’t give you any options to manually tell the stupid thing the blatantly obvious. So, no upgrade options
  • managed to break the code that allows you to edit “table of contents” in word documents, so that trying to load and edit a standard word doc crashes the whole app.

Oh, and if you install none of the office apps, it still takes up 190Mb disk space. Who says open-source software isn’t untested bloated crud? :frowning:

Not impressed :(.

#2

Works fine for me, here, but then I didn’t do compilcated installation procedures… :slight_smile:

Btw. bugs are always there in software. OpenOffice is in no way more buggy than commercial software.

#3

There’s a standard procedure known as “regression testing” to prevent all of the problems I had, except for the contents one (which could be due to something else particular to each of the files I’m loading, although my guess would be that it’s unlikely).

You should never release a minor update without checking that it works at least as well as the previous one did. MS never really got the hang of this either :frowning: (ability of service packs to re-create previously-fixed bugs).

#4

I know it’s very annoying when you want to use a new version of a software and it doesn’t do what it should. :frowning:

[quote]MS never really got the hang of this either :frowning: (ability of service packs to re-create previously-fixed bugs).
[/quote]
That’s right. And there are even more well known software companies who didn’t get it either (however with MS is so annoying because soooooooo many people have to deal with their bugs)…

#5

I want it all for free, to run perfectly and I want it now ::slight_smile:

#6

[quote]I want it all for free, to run perfectly and I want it now ::slight_smile:
[/quote]
Then use TeX as your word processor :slight_smile: And read the comments in the source code where Knuth offers money for bugs found (he actually did pay). I believe bug-free software is entirely possible and practical - in some contexts.

#7

True, but in his topic title he named this a typical of open source, in which case I present the typical open source software user… a person who wants it all for free, to be perfect and right now.

Both are a generalisation as there are plenty good projects that are very throuroughly tested (or at least have some automated unit/regression testing going on)

#8

I think the topic would be better renamed to:

Software …typical

;D

#9

All good points, IMHO.

The only reason I get so mad at OO here is because these are regressions - they’ve built up an expectation that the product would work, to the extent that I’ve then become reliant upon it, and then it suddenly stops working, and it’s like having the carpet pulled out from under you.

The only comment above I take issue with is the one about “wanting it all, for free”. In one sense, it’s a good (rhetorical) point - you pay nothing, why should you expect anything?

On the other, if you knowingly and deliberately offer a service or product, and even encourage people to use it, that also places moral responsibilities upon you, as a provider. (This, of course, is part of why you see software always come with those sometimes illegal disclaimers - this moral obligation is so engrained it’s actually enshrined in law. And yes, I do think it utterly outrageous that companies get away with disclaiming things that in fact it’s sometimes illegal to disclaim, just because their users don’t have the money to employ a lawyer to challenge it!).

These two sides to the coin have no one right answer, but on a case-by-case basis I feel that if you are developing something you know is being used by hundreds of thousands of people to do their daily work, and you release a non-beta update, the obligation outweighs the “get what you pay for”. There are other cases where I’d say the latter outweighs the former.

#10

[quote]I believe bug-free software is entirely possible and practical - in some contexts.
[/quote]
Yes, in the “Hello World” context.
:slight_smile:
In others, not. Simply because man isn’t error free, so will be his products.

Of course you can do a lot to make the probability of errors very small. Think at airplanes… (If I remember right I read that usually 3 different software systems (trying to calculate the same) run on 3 different hardwares and if one comes to another result than the others, they do a 2/3 voting to find out which one probably is wrong.)

Oh, off topic…

#11

Please rename the topic, if you agree, that your post is not really related to Open Source. :slight_smile:

#12

[quote]All good points, IMHO.

The only reason I get so mad at OO here is because these are regressions - they’ve built up an expectation that the product would work, to the extent that I’ve then become reliant upon it, and then it suddenly stops working, and it’s like having the carpet pulled out from under you.

The only comment above I take issue with is the one about “wanting it all, for free”. In one sense, it’s a good (rhetorical) point - you pay nothing, why should you expect anything?

On the other, if you knowingly and deliberately offer a service or product, and even encourage people to use it, that also places moral responsibilities upon you, as a provider. (This, of course, is part of why you see software always come with those sometimes illegal disclaimers - this moral obligation is so engrained it’s actually enshrined in law. And yes, I do think it utterly outrageous that companies get away with disclaiming things that in fact it’s sometimes illegal to disclaim, just because their users don’t have the money to employ a lawyer to challenge it!).

These two sides to the coin have no one right answer, but on a case-by-case basis I feel that if you are developing something you know is being used by hundreds of thousands of people to do their daily work, and you release a non-beta update, the obligation outweighs the “get what you pay for”. There are other cases where I’d say the latter outweighs the former.
[/quote]
That would be me :slight_smile:

Don’t get me wrong, I believe in delivering the best possible whether it is closed or open source, but my general understanding of open source users is how they always expect perfection for free right away.

The source is open, if it bothers you so much, contribute.
Though I do agree that regression bugs should not happen, especialy since everybody knows regression testing can be simple.
However, having picked up some open source work, I know how these “not fun” things hardly get picked up

#13

No, I think it is possible for complex systems (I give the example of TeX) to be bug free. Just because people make errors doesn’t mean those errors can never be fixed. Errors are finite, I see no reason why it can’t be possible to fix ALL of the errors. If you stop adding new features so you can actually reach a point where the software is “done” then you will have something with a fixed number of errors. If you then work to fix the errors, introducing new errors at a rate lower than the rate you fix existing errors, it is possible to fix all the errors.
This is assuming that there is not a fundamental problem with the entire system such that the design requires that some error not be fixed without forcing another error elsewhere, so the two errors can never be fixed at the same time.

#14

But what sort of organization is actually able to stop adding new features for long enough that the bugs actually get fixed? Most software companies seem to feel that if they stopped the flood of new features for that long, their competitors would eat them.

#15

[quote]But what sort of organization is actually able to stop adding new features for long enough that the bugs actually get fixed?
[/quote]
It seems that Open Source projects are exactly the sort that can do this, since they generally aren’t for profit.
Some things can be “complete” that is you design a thing to do X, you get it to do X, and it’s done. Even in the case where it is only “done” for a few months until you need to upgrade something, there is still a point where you say - this is all that version 1.0 is - no new features will be added to the 1.x stream. You could make a bug free version 1.n that added no new features, while introducing new (and buggy) features with 2.0.

BTW, I don’t mean to imply that any of this is easy. I just think it is possible. The notion that software must have bugs is something I simply don’t believe.

#16

Indeed; I actually use several open-source products (usually quite small things, only a handful of classes) that have only ever had one major release and are no longer developed - because there are no outstanding bugs.

I know of cases where I’ve written bug-free software for use in a commercial environment, limited only by the bugginess of outside factors (like windows, or a new JVM release with a regression bug). It is possible, and quite often commercially viable, to sufficiently analyse software that you know it has no bugs of your making. Of course, that excludes a lot of bugs from the equation, especially with complex systems (e.g. in a heterogeneous environment, development is hard precisely BECAUSE you are having to write more workarounds for other systems’ bugs than you are writing fixes for your own bugs!). Also, just because it’s commercially “viable” doesn’t usually mean it’s also commercially “advantageous”.

I concur with SWP that open-source has (in one way) a fundamentally easier time getting to bug-free software - but IME it’s generally of no better quality than closed source software. Which is a little sad, but also, I suspect, a reflection of the fact that most of the best programmers are paid alot of money to work hard enough not to have the time to do opensource programming :(; people like the JBoss authors are the notable exception in opensource projects (e.g. I can list os projects off the top of my head with top-notch developers, and if I could even recognise a lot of their names - that’s how much they stick out in your mind). Which is not to say there aren’t a large number of them in absolute terms, but in percentage terms it often seems as though it’s practically nothing :(.

OTOH, early Apache authors didn’t have a clue what they were doing (hence the name! …And the stonkingly poor performance!), and early Tomcat authors should be shot .

EDIT: of course, these days some of the work done by the Apache groups (jakarta) is excellent.

#17

Nevertheless they seem as prone to being seduced by new features as many commercial organisations. At the risk of being ‘on topic’, just look at the endless list of things that people would like added to Java.