iraf.net

When will you get a Windows release of IRAF?

I wouldn't put it at the top of the list, despite the Windows monopoly there hasn't actually been a lot of demand for a Windows port over the years. Chris Burke at Ohio State has released a Cygwin port (just Google "iraf cygwin") that appears to work, and I did one a few years back but never released because of the potential support nightmare. Even so, you need to first install Cygwin, all
the X stuff, compilers, etc, it is almost easier to dual-boot linux or run VMware
instead.A native Windows port is technically possible now, save for the need for a suitable graphics terminal not X-based. Overall, a lot of work. I'm interested to see your poll results though, but I suspect in a head-to-head vote MacIntel would win.-Mike

[quote:e317faed87="fitz"]I wouldn't put it at the top of the list, despite the Windows monopoly there hasn't actually been a lot of demand for a Windows port over the years. Chris Burke at Ohio State has released a Cygwin port (just Google "iraf cygwin") that appears to work, and I did one a few years back but never released because of the potential support nightmare. Even so, you need to first install Cygwin, all
the X stuff, compilers, etc, it is almost easier to dual-boot linux or run VMware
instead.A native Windows port is technically possible now, save for the need for a suitable graphics terminal not X-based. Overall, a lot of work. I'm interested to see your poll results though, but I suspect in a head-to-head vote MacIntel would win.-Mike[/quote:e317faed87]Mike,
more than a win port, a macIntel port, a xxx port... and so on, I'd like to see a "GNU style" port which "automagically" compiles (i.e. ./configure && make && make install) on the various unix/posix platforms (for example linux/amd64 and then, yes, even win/cygwin).... It would be a lot more useful since many unix flavours supported by iraf are almost dead... Can we help to make such port? It would be interesting to a have not only an "installation guide" and a "site administration" guide, but also an "internals/develor" guide to explain how Iraf works from the inside (the core of the system, not the various astronomical-application-specific tasks) to leverge the free-software community to develop iraf (imagine an Iraf sourceforge project... :wink: )Many thanks, and best regards.
Emiliano Gregori

[quote:c3d7c9d792]more than a win port, a macIntel port, a xxx port... and so on, I'd like to see a "GNU style" port which "automagically" compiles (i.e. ./configure && make && make install) on the various unix/posix platforms (for example linux/amd64 and then, yes, even win/cygwin)....[/quote:c3d7c9d792] More Debian philosophy, huh? If a port exists then "./configure" basically breaks down to resetting the package architecture and a "make" can be done using the existing build scripts (or some variation). The number of people who would actually build from source in this way is relatively small, the amount of work to create a real autoconf and Makefiles is HUGE, and given the payoff I'm not sure this should be the focus right now. Certainly building a "sysgen" script is easy enough and something we might do with an "IRAFNET" release of the system; Debian zealots can go from there to meet their build requirements, once they're done arguing about the licenses 8-)[quote:c3d7c9d792]It would be a lot more useful since many unix flavours supported by iraf are almost dead...[/quote:c3d7c9d792]Effectively we're down to just Suns and PCs, but even that makes up more than half a dozen platforms. I doubt this site will be much involved in HPUX
support, but we also shouldn't be in a position where new linux releases are
forcing us to an IRAF release on the same schedule! Binary distributions are what most people want/use, and given the (even more) limited resources are what we should focus on now.Certainly contributions along the lines you suggest are welcome, and more than we've gotten in the past.[quote:c3d7c9d792]Can we help to make such port? It would be interesting to a have not only an "installation guide" and a "site administration" guide, but also an "internals/develor" guide to explain how Iraf works from the inside (the core of the system, not the various astronomical-application-specific tasks) to leverge the free-software community to develop iraf (imagine an Iraf sourceforge project... [/quote:c3d7c9d792]Is that a volunteer I hear? There are probably a variety of documents still needed in the system, both for users and developers. There are also a lot documents already in the community, although not all of them accurate. A good place to start would be by adding links here to the various department tutorials and intro guides already available (just Google 'em) and then editing for accuracy. Another approach would be something like a Wikipedia of IRAF bits similarly edited.Writing documentation from scratch takes time and while a complete brain-dump of what we know of how the system works might be useful, the audience might not be large so it becomes a question of priorities again (for us). Anyone interested in projects like this should start a thread in the General group looking for partners, or just start working and submit it when done. We'll certainly help in any way we can, but remember this is now a community-based system, not simply a new venue for making requests.-Mike

Hello Mike, and thanks for your reply!IMHO the problem is the following: if you (as a manteiner) and us (as users, programmers, hackers, and so on) want IRAF to survive for long time wih the fast evolution of unices and hardware platforms, IRAF source should become the most compliant as possible with "de facfto" programming standars. With programming standards I mean: compilers (for example: gcc), libraries (posix enviroinment...), programming models (multithreading, MPI, and so on...), building tools and packaging systems (well, this seems the real nightmare...)[quote:ba33e5edf9="fitz"] More Debian philosophy, huh? If a port exists then "./configure" basically breaks down to resetting the package architecture and a "make" can be done using the existing build scripts (or some variation).
[The number of people who would actually build from source in this way is relatively small, the amount of work to create a real autoconf and Makefiles is HUGE, and given the payoff I'm not sure this should be the focus right now.
Certainly building a "sysgen" script is easy enough and something we might do with an "IRAFNET" release of the system; Debian zealots can go from there to meet their build requirements, once they're done arguing about the licenses 8-)
[/quote:ba33e5edf9]I agree with you that a different autoconfigure script for each port will be totally unuseful.
But once you have ONE clean and portable source (not a collection of ports), you can issue the faomous "./configure && make && make install" in all your develpment machines and build the binary ports (present and future ports) and distribute them on the IRAF site (in this way we could speak of a "debian style"). If a brave user would like to run IRAF on his exotic hardware (maybe the IBM Cell ?! ), he could download the source and try to compile, provided he has the right (standard) development tools! (well... I admit it sounds a little bit utopian... but I see it as the only way to a true "free and reusable" software)I'm not the mean debian zealot (dissussing about licences all the time), or the mean apple zealot (talking about PowerPC vs. Intel..... you know PPC rules :wink: ), but I'm interested in using IRAF and keep it alive, beacuse it contains a lot of valuable code![quote:ba33e5edf9="fitz"]
Effectively we're down to just Suns and PCs, but even that makes up more than half a dozen platforms. I doubt this site will be much involved in HPUX
support, but we also shouldn't be in a position where new linux releases are
forcing us to an IRAF release on the same schedule! Binary distributions are what most people want/use, and given the (even more) limited resources are what we should focus on now.
[/quote:ba33e5edf9]In my opinion IRAF is now the perfect example for the famous "Cathedral and Bazaar" paper: up to know IRAF has been developed very well in the NOAO cathedral... maybe because when you invented IRAF it was the only possible choice due to the big intrinsic differences of unices (HP vs Sun vs Irix and so on..), but know the world is a big bazaar full of cheap and powerful linux (or bsd) boxes... It's true that a new distro often breaks an old (IRAF) binary package, but the probability that a new linux kernel breaks a "GNU compliant" source code is little, and if breaks it someone in the "bazaar community" will fix it.

[quote:ba33e5edf9="fitz"]
Certainly contributions along the lines you suggest are welcome, and more than we've gotten in the past.
Is that a volunteer I hear?
[/quote:ba33e5edf9]I'll do my best!
[quote:ba33e5edf9="fitz"]
There are probably a variety of documents still needed in the system, both for users and developers. There are also a lot documents already in the community, although not all of them accurate. A good place to start would be by adding links here to the various department tutorials and intro guides already available (just Google 'em) and then editing for accuracy. Another approach would be something like a Wikipedia of IRAF bits similarly edited.
Writing documentation from scratch takes time and while a complete brain-dump of what we know of how the system works might be useful, the audience might not be large so it becomes a question of priorities again (for us).
[/quote:ba33e5edf9]Maybe a very important document to publish could be a "How to Compile IRAF" guide. I'll try to search on Google documents about IRAF internals, and I'll post links and other interesting things. A WikiIRAF open to community users and developers could be a great tool![quote:ba33e5edf9="fitz"]
Anyone interested in projects like this should start a thread in the General group looking for partners, or just start working and submit it when done. We'll certainly help in any way we can, but remember this is now a community-based system, not simply a new venue for making requests.
-Mike[/quote:ba33e5edf9]Many thanks for your great work!
Emiliano

I'll volunteer to help IRAF too!

[quote:0dbc5ca23a]IMHO the problem is the following: if you (as a manteiner) and us (as users, programmers, hackers, and so on) want IRAF to survive for long time wih the fast evolution of unices and hardware platforms, IRAF source should become the most compliant as possible with "de facfto" programming standars. With programming standards I mean: compilers (for example: gcc), libraries (posix enviroinment...), programming models (multithreading, MPI, and so on...), building tools and packaging systems (well, this seems the real nightmare...)
[/quote:0dbc5ca23a]In this business, the 20+ years that IRAF has [i:0dbc5ca23a][b:0dbc5ca23a]already[/b:0dbc5ca23a] [/i:0dbc5ca23a] been around should qualify as a "long time". The original system architecture remains intact and I would argue was "object oriented" long before the phrase came into vogue. It has allowed us to port a million line of code to more than twenty different OS's over the years by simply porting the kernel and XC compiler, MKPKG has been dead solid compared to Makefiles, the Imakefiles that were supposed to fix those problems, and autoconf which is being replaced by Ant as it moves out of Java space. GCC and Posix standards are already used in the system (have been for many years), MPI requires applications changes and build tools/packaging systems are a very small part of the user base.If by your statements you mean we need to be more open to other programming/scripting languages, need to be more agile to changes in linux, and things are a little dated then I certainly agree. However, I doubt that IRAF is "dying" because people can't build from source easily. Source-only systems like Debain/Gentoo are a hacker's niche platform, easier installation (by packagers or our own means) will do much more to help the majority of users than the (current) "de facto" standards you mention.[quote:0dbc5ca23a]I agree with you that a different autoconfigure script for each port will be totally unuseful. But once you have ONE clean and portable source (not a collection of ports), you can issue the faomous "./configure && make && make install" in all your develpment machines and build the binary ports (present and future ports) and distribute them on the IRAF site (in this way we could speak of a "debian style").[/quote:0dbc5ca23a]No argument here, I'll talk to you again in a year or so when you're done with the uber-script to build the system 8-) PC-IRAF is one source that compiles on a half-dozen platforms, it just doesn't use the infamous command to do it. So? I remember when "xmkmf && make World" was all you ever needed to know.[quote:0dbc5ca23a]In my opinion IRAF is now the perfect example for the famous "Cathedral and Bazaar" paper: up to know IRAF has been developed very well in the NOAO cathedral... maybe because when you invented IRAF it was the only possible choice due to the big intrinsic differences of unices (HP vs Sun vs Irix and so on..), but know the world is a big bazaar full of cheap and powerful linux (or bsd) boxes... It's true that a new distro often breaks an old (IRAF) binary package, but the probability that a new linux kernel breaks a "GNU compliant" source code is little, and if breaks it someone in the "bazaar community" will fix it. [/quote:0dbc5ca23a]NOAO as a Cathedral is amusing on various levels. Expecting that [b:0dbc5ca23a]now[/b:0dbc5ca23a] we will suddenly have people stepping up to fix things is also amusing -- Source for IRAF (et al) has been around since the beginning of the project and while expecting part-timers to hack SPP might be expecting too much, where have all the the bazaar community been over the years to "fix" the many complaints about the CL or 8-bit-only XImtool? These are C/X11 programs that shouldn't lack any expertise in the community to fix.. I'd much rather see a patch for a 24-bit XImtool than an autoconf script, and given limited time to work on either which do [b:0dbc5ca23a]you[/b:0dbc5ca23a] think is more important?By far IRAF (as with most projects) has a "user community" and not a "developer community".. I certainly hope that can change, but until it does those who [b:0dbc5ca23a]can [/b:0dbc5ca23a] develop should focus on the users first, and then ourselves. (This applies not only to IRAF software but to this site as well, e.g. in your suggestion for new documents or my suggested projects or simply adding links to IRAF resources on the net.) If that means writing an application in SPP/CL then so be it, major changes require a major effort and we don't (yet) have the resources for all the major efforts needed. Your suggestions about new documentation are good ones, as are your system ideas, but I'd urge you to keep some perspective about what can realistically be done, and what should be done first.-Mike

In my opinion iraf have to be rewrote totally from scratch with a totally different methodology. Bye.P.D. Why don't you redistribute IRAF from your site? Maybe the license problems are not only for Debian hackers? And... CORRECT THE COPYRIGHT INFORMATIONS IN FAQs, please!

[quote:9c9b8e3a64]In my opinion iraf have to be rewrote totally from scratch with a totally different methodology.[/quote:9c9b8e3a64] Good luck with that.[quote:9c9b8e3a64]Why don't you redistribute IRAF from your site? Maybe the license problems are not only for Debian hackers?[/quote:9c9b8e3a64] Once we have something different to distribute, we probably will.[quote:9c9b8e3a64]And... CORRECT THE COPYRIGHT INFORMATIONS IN FAQs, please![/quote:9c9b8e3a64]The FAQ does say it is only "part" of the copyright, but I added a line saying where to find the full notice.