I've been watching the Lazarus and FPC mailing list for about a year, trying to decide whether to give these fine systems a go. But every time I get close to seriously trying them out I shy away. That is because I see a new raft of messages telling stories of how an installation and/or upgrade failed for some inexplicable reason. I just don't need that kind of hassle in my life.

Sometimes it is a simple install that results in a configuration that cannot compile a simple demo program. Sometimes it is an attempt to upgrade to a new compiler that results in previously working programs failing to compile. The description given usually seems like the person did a reasonable thing but they nevertheless end up with an error that they cannot explain and cannot easily fix.

The regularity of these types of messages is a phenomenon in itself. I cannot myself judge whether these are unusual results. I wonder how many people download the packages, install, and go on their merry way with no problems.

So, I'm asking everyone to discuss this a bit so we can develop an understanding of what is going on here. I'm not criticizing the current way of doing things and I don't want to jump right to "solutions" to specific favorite annoyances. Rather I'd like to assess the overall situation and determine where the problems lie.

Is the build system flawed? Is the installation procedure flawed? Are the upgrade procedures flawed?

Is the documentation for all the above flawed?

Are the main problems specific to certain platforms?

Would it be a worthwhile goal be to make the install and upgrade procedures simpler? Are they already as simple as they can be?

Are there more problems with Lazarus than for FPC itself? Could improved integration of Lazarus and FPC help?

Does the core team think the situation is "not all that bad"? (I often get that impression.)

Is there already an effort planned or underway to improve things? (I have not seen one beyond a couple of individuals making their own installer.)

What do you think?

Doug Chamberlin wrote

What do you think?

Honestly: you exaggerate the problem, IMHO. I'm not saying that bits cannot be improved here and there, but...

For every single panicked bugreport, there are 100 people who just installed everything in 5 mins and are using everything without problems. These people don't flood the mailing list with messages that "it's was trivial to compile and install FPC and Lazarus from sources if you know what you're doing", so you don't see them.

I would advise to try and install Lazarus for yourself :)

Michalis

Michalis Kamburelis wrote

Honestly: you exaggerate the problem, IMHO.

I don't think its an exaggerating to observe the regular reports of install and upgrade problems. Just look at any 1 month period over the past year or so.

For every single panicked bugreport, there are 100 people...

There are no doubt more successes than problems. But I don't see the problem rate going down in any significant way. Nor do I see a reduction in problems when people do unusual, but reasonable, things. In short, I think the system is more brittle than it should be. But that's just my impression.

So I think you are guessing here. Do you have any real numbers?

Every piece of software has bugreports about install/upgrade problems. No matter how easy the procedure is. Partially, that's because the environment constantly changes --- new versions of operating systems, new versions of libraries, etc. This is a compiler and an IDE, tied to a lot of different libraries and dependencies, on a lot of different systems. And people are using it in a myriad of different ways.

My main point was: drawing from this a conclusion that there's some general problem, needing some single grand solution, is an exaggeration. There are just many small problems, that are being eliminated, one by one, by developers. And in the meantime, new OSes and libraries appear, sometimes introducing new problems...

I don't have any real numbers. My "100" was of course just a guess. Although you can look at the number of downloads e.g. on SourceForge. This week, https://sourceforge.net/projects/freepascal/ says 12 thousands downloads. https://sourceforge.net/projects/lazarus/ shows 10 thousands of downloads. That's quite a lot, and it's understandable that some percent of these people stumble on some problems.

The only real data that I have is about myself. I've been installing FPC since around version 1.0.2, on Linux, Windows, and later Mac OS X. And I don't remember having to struggle with something terrible. I'm installing Lazarus usually by compiling it with whatever FPC version happens to be currently stable --- and it also pretty much just works.

Hence, my simple advice --- first try to install FPC/Lazarus yourself, and see if there's really a problem. You may as well find that it's as easy as running an installer and clicking "Forward" until the installation finishes :)

Michalis

I was not claiming there was a big problem requiring a grand solution. it's just that there is no other software I have ever used that has as many recurring, similar trouble reports over such a long time period. That bothers me.

So I was just commenting on the persistent reports of trouble and asking for a group assessment/comment/consensus as to how big a problem there is. Sounds like your opinion is that there is no real problem here, move on. I disagree. I think there is a distinct problem. Perhaps it is endemic to a complicated cross-platform, cross-compile tool. Or from having two tools that must work well together but cannot merge into one project. Don't know.

I have installed Lazarus and FPC several times over the past few years. So, urging me to do that is completely unnecessary and missing my points. My points are not about *my* experience with the software or solving a problem I have had with it. They are examining the health of the whole Lazarus/FPC ecosystem and asking questions I have not heard asked before.