You want me to adapt the Allegro API? In which case I can hardly claim to be writing an Allegro driver, surely? I'm thinking of some of the things I usually mention which I've found desperately useful but which other people seem able to dismiss out of hand - e.g. overlay surfaces (which are coming in Allegro 5) and gamma ramps (which weren't, last time I checked).

Thats true, and something I had not taken into account. Although it would be impossible to argue that Allegro was 'more popular' (i.e. more used) amongst companies, I must admit that my statement that "SDL is probably more popular" was based on fuzzy information sources such as the number of web sites, the number of posts here versus the SDL mailing list / usenet group. Also, I assume that if you find yourself in a situation where SDL comes with your compiler, you will almost certainly turn to SDL by default, meaning that increased popularity amongst companies filters down to increased popularity amongst users.

In terms of the sort of projects that used to make it big so that I noticed from the Allegro camp - emulators and remakes - the former market has more or less reached critical mass and the latter has been eaten away by Blitz and Dark BASICs.

In the short term Allegro is no doubt loosing out due to its horrible API, but on this topic I am towing the party line and of course, we are all aware that version 5 will fix the problem.

You'll agree though, that there is a line between providing tools for a minority and trying to create a minority? How about a new poll on the front page :

"
Which function naming scheme would you prefer for Allegro 5?

Another problem with the developers is that they seem to assume they can state the opinion of the Allegro users without actually asking them. I'm not expecting such a poll to deliver them any sort of sudden shock, and only a fool would expect them to stage a U turn if it does, but since you don't seem to have much else to put in the polls do you not agree that it would be very healthy to use it as a simple forum to allow the expression of wishes for the new API by the majority of users where the developers cannot simply pick us off one by one?

This is another area I hadn't thought too hard about. I would be perfectly happy to, and am abstractly capable of, developing an SDL target for Allegro, but for two problems :

I am aware that the reason Allegro does not build on MSVC is that it still has a large quantity of AT&T type assembly which MSVC does not support. I am also aware that the developers have at some stage taken on board that switching to intel syntax assembly for version 5 would not limit target platforms in a world where NASM exists, so I assume that will happen?

I agree. What we need is a formal, public, convenient means to openly discuss these things. The mailing list is hell for dialup people such as me, and has the disadvantage of not having a means to get any sort of consensus amongst anything like a good cross section of the Allegro user base. Could we not have some sort of forum here where people can post ideas, give the developers (or whoever) the right to reply for a short while, then move to a vote on the issue if required?

I wouldn't call it 'horrible', but 'inconsistant' and suffering a bit from legacy. Part of what makes Allegro nice is that it's API is simple. If all the functions were renamed, the Allegro 4 API would actually be pretty nice. (Yes, it still wouldn't support some things that today are nice to have, but it would be nice nonetheless.)

Actually I'm currently working on a survey system for a commercial website I'm doing, and I plan on making it available on Allegro.cc so that certain developers can set up surveys and ask a bunch of questions at once, instead of a limited fashion from what the polls provide. At that point, I/they could take suggestions for questions to ask. Doing it in a survey fashion would allow the results to be easily believed (ie, only one response per member) and tallied.

The majority of Windows users, no doubt. But, can one even build SDL on Windows/MSVC? Last I checked, I thought the SDL build process (on Windows) looked much more complex than Allegro's.

A few points:

The underscore thing I really don't understand. The idea that one library should use a naming convention because another one uses it is surely fatuous.

Please don't have voting for the allegro development process. By all means gather opinion, but doesn't experience show that you need a dedicated person or small core group to really deliver a great product (think Linus Torvalds, Wozniak/Jobs, Shawn Hargreaves, Sam Latinga if you will, or even Ma****ew Le****on )

As for the build process, what's the process for SDL/Linux? Or MinGW? One benefit of Allegro is that if you use a free cross-platform compiler (i.e. gcc, mingw, or djgpp) to build a free cross-platform library, the process is almost identical.

Lastly, I'll say it again - a focussed method to get opinion on development questions and a way to feed back the decision would be great.

Pete

For all that is good and holy, keep it at alternating_underscores. The other schemes actually give me a headache (and yes, I used them in the past).

If this is such a big issue, the solution is simple: provide a wrapper header consisting of static inline functions that changes from the one to the other. Do you know I seriously considered doing that for OpenGL because MixedCaps conflicts badly with my own coding style.

Just voicing my opinion, of course.

Note that SuSE Linux comes with Allegro on the CD. I'm not sure, but I think it doesn't include SDL. I could be wrong though.

erm... [AD] generates less than 10 messages per day on average, I think. I'm on dialup and find the mailing list far more convenient than these forums, for instance. I'm not sure why it's hell for you.

Well, there is the Allegro development forum? I'm not sure how many of the developers read the forums, but I see Bob, Vincent, Peter (Wang) and Grzegorz around here regularly at least.

Oh, and I agree with Peter Hull: voting would be a bad idea IMHO.

What!? Personally Im not a allegro developer, But I deffinitly prefer the underscores... gives more contrast than the CUpperCaps method. (or worse yet, szv_CUperCapsBlahJoe)

What is ugly to one person wont be to another. I find that UpperCaps crap fugly as he11. But thats my OPINION.

Want lots of examples where underscores and no caps are used? libc, libstdc++, libvorbis, libogg, libvorbisfile, lua (Im sure theres a TON more...). Now tell me that the majority of APIs that (PC) game developers come into contact with don't use underscores. heh. more like APIs that are DX or based on DX. hmmm... weird no?

[edit]

what? for allegro I just download Mingw v1 and everything is peachey. why msys? and make should be make... not this mingw32-make stuff.

Now if youre thinking, oh but my MSVC or borland or whatever make is going to conflict with it!!! that means you havent setup your system correctly. you setup a runner batchfile for each compiler and away you go.

[edit2]

Wana know what? screw this... The developers are doing it all in thier spare time. you dont like a decision? you have two (valid) options. You can ditch allegro, or you can do something about it. Even if that means forking allegro, go right ahead. uggghhh. Quit making such a fuss.

I already did.

Oh, yeah, and APIs which are OpenGL or based on OpenGL too. If you like I could carry out a statistical test by heading down to my local game shop and working out the proportion of games based on either DX or GL versus the proportion not based on either of those.

Indeed a long list. But I notice that the one library I know well from your list doesn't actually use underscores at all, so you'll excuse me if I don't take your word as bible. Or perhaps you can point out the underscores in :

zlibVersion, deflateInit, deflate, deflateEnd, inflate, inflateEnd, deflateInit2, deflateSetDictionary, deflateCopy, deflateReset, deflateParams, inflateInit2, inflateSetDictionary, inflateSync, inflateReset, compress, compress2, uncompress, gzopen, gzdopen, gzsetparams, gzread, gzwrite, gzprintf, gzputs, gzgets, gzputc, gzgetc, gzflush, gzseek, gzrewind, gztell, gzeof, gzclose, gzerror, adler32, crc32

It looks to me like they've stuck to the stdio naming convention for functions which are pretty much stdio equivalents and branched into capitalisation elsewhere. Mind you, TF Sees underscores in libc, so perhaps that is what was confusing you?

Well, off hand I'd have to say that I've never heard of anyone using much of BOOST before, but to claim as such was truth would be to fall into the trap affecting some of the developers : assuming I know things for which I have no evidence.

I don't see why this is a relevant question in a thread principly about why SDL seems to be more popular than Allegro, and using underscores as a particular example of how Allegro differs from the sort of thing that would make it more popular. Unless you want this answer :

But I fail to see how you won't have picked that up as my opinion already.

It isn't, but why do you ask? And who suggested it was?

I notice that you draw a distinction between 'doing something about it' and 'making such a fuss', which shows you probably haven't realised that most changes in this world come about only when somebody makes a fuss. And if you think that arguing for a better source of information from which the developers can digest the opinion of users is not helping them then I think you might want to address that belief also.

I am unwilling to enter into personal attacks, and would kindly request that you assume the same position. Directly answering the question, yes I did. And I overwhelmingly saw functions such as :

tolower, fopen, cfsetspeed, memcpy, pclose, printf, setgid, signal, umask, wordexp

And only very few functions such as :

va_end, fgetpwent_r, inet_ntop

Which don't even conform to the same underscore system as Allegro anyway. Did you post the link you intended to?

I answered the question asked by Peter Hull. I have this bad habit of doing such things.

I hope you can understand why I mistakenly took the suggestion that I had not bothered to read the links and furthermore that I was almost certain not to as a personal attach. Apologies.

Indeed I now realise that you have a point in that instance, however I would argue that you would not have a point were you to instead argue that Allegro takes its cue from such things as libc, since to my mind there is a gulf of difference between the two naming conventions. Also, I must question why you then attacked whether I'd seen the link or not when I clearly stated that I didn't see how you saw underscores in libc, when surely you wanted to attack me for not having properly digested your post?

Speaking of which, much as I enjoy talking about naming conventions, and making it explicit that I am not shying away from my belief that Allegro is the 'odd one out' in this respect amongst popular game related libraries, how about we return to a conversation about the possibilities of finding a better way to gauge opinion amongst Allegro users for the benefit of the developers?

Do people agree or disagree that currently the developers seeming to just assume they know what people want is not necessarily going to produce the best output?

I repeat that I favour the idea of polls that could be opened on development 'hot topics', so that the average user could more easily and fairly express their opinion, and more users could be conveniently surveyed than via mailing lists. Obviously there is absolutely no binding reason why the developers need ever look at or take part in these polls, but it would provide a useful statistical argument for those who do actively take part in development discussions.

I'd just like to say:
STOP IT NOW
Please, Thomas F and Thomas H, do you really think that underscores or not underscores should be the priority for Allegro development?

Pete

ps. there are underscores in SDL, aren't there sorry sorry sorry

Dose it realy matter wever youTypeLikeThis or like_this? I think that if it dose then you need to get out abit more(e.g. get a life) or like Evert said ...

I can understand that you want your code to look good so should any self respecting programer but if its good code then the proth is in the binary. And lets face it even if you write shaby code as long as people like it and can play your games its good.

Anyway to sum up Allegro vs. SDL round 1200:
Try both! Make your own mind up.

If your a biginer Allegro is sugested as many of us managed to start and learn C while messing around with Allegro.

If your a person that likes good looking code then it realy is a personal thing as some ppl like:
ALDrawBitmap
al_draw_bitmap
AlDrAwBiTmAp <--- i think this is realy artistic but dumb as hell.

Don't take it personal If I'm arguing, I'm arguing about what you say (and perhaps the way you say it), not you personally.

There's no such thing now, so from that perspective having anything is better than nothing. It's a long road to a native MSVC build anyway.
The main issue I can think of with switching to seperate assembler code for MSVC is that there are two assembler code bases for one platform. I imagine this will be a source of discussion yet.

al_blit has a tendency to be mis-typed as al_Blit, and is more keystrokes than alBlit, as Marcello said; but yeah, this really don't make that much difference.

Anyway, I realize what you meant about MSVC; for some reason I was thinking build FOR msvc, not build WITH, and, um, I dunno; I'm fine now
It would be cool to see a better build process for allegro & msvc; yeah, I guess it would.

It's cool that Matthew intends to set up something like that ... maybe we'll see some interesting data when he does! Yay to Matthew!

And to Evert: the wrapper idea thing sounds cool! Heh, and we'd have one less silly argument ...