From: John V. <ja...@gm...> - 2005-09-25 23:48:57
|
On 9/26/05, Brandon J. Van Every <bva...@gm...> wrote: > amores perros wrote: > > >> From: "Brandon J. Van Every" > >> > >>>> From: "Brandon J. Van Every" Reply-To: mailing list > >>>> > >>>> Since the installer is therefore pointless, shouldn't it be removed > >>>> from the download area temporarily? Or does it work on some > >>>> systems and not on others? > >>> > >>> > >>> It may serve the purpose of driving away those who are not > >>> sufficiently committed to continue trying after wasting hours > >>> downloading the 20+ tarballs, only to have the installer > >>> delete them all -- so that it separates the newbies into the > >>> groups of those who are either on broadband, and willing > >>> to continue after the main installer fails, or who are on > >>> dialup and REALLY committed and willing to spend > >>> hours more, from the newbies who lack the gumption > >>> to push through the apparent brokenness in pursuit > >>> of hope. > >>> > >> Acknowledging that we're discussing a moot issue, I will ask: do you > >> think that's a good thing or a bad thing? Also, do you think it's > >> consistent with attempting to provide installation software in the > >> first place? > >> > Again, acknowledging that this discussion is academic, I'm not convinced that this discussion is academic - another installer discussion is not what we need. This is a very visible problem, and will continue to be until a working installer(s) is available. The current installer was rushed through the phases (a few Snapshots and then a Release [1]) because it addressed very real needs that the previous installer didn't cater for. Earnie did a good job on the current installer, but from what I understand, there was a dependence on a closed-source binary [2] and non-free (beer) toolchain (Borland Delphi [3]), and so that installer hit a brick wall. I think that we should not leave the situation as it is until the next installer is available, as it will put pressure on Dave Murphy's replacement. A new installer should go through the release system slowly, so that it has sufficient time to knock of any weird bugs that may arise, and also so that the MinGW community has time to make sure the next installer meets all needs [4]. I am looking forward to an installer that is built on free tools [5]. > > until such point as there is a new installer which actually > > works. On the negative side, it may make mingw look broken, > > "May?" It does, most thoroughly. Realize, MinGW exists in a Windows > world, where working installers are taken for granted. Of course there > are 2 cultures being served here: Windows users who expect things to be > plug-and-play, and Unix users who feel themselves stuck with Windows > development and don't want their code GPLed. (In my experience > -mno_cygwin is a non-option. Open source libraries tend to be deeply > stacked and -mno_cygwin tends to break them.) I agree that the current installer reflects badly on MinGW, which is categorised as "6. Mature" on the project page. The downloads section should recommend the old installer [2] until the new installer has undergone significant testing and there is consensus that it is now suitable for everyones needs. The current ISTool installer should be moved back to Proposed. 1. http://mingw.org/news.shtml 2. http://www.istool.org/ 3. http://www.jrsoftware.org/isinfo.php#features 4. http://lists.zerezo.com/mingw-users/msg00469.html 5. http://cvs.sourceforge.net/viewcvs.py/nsis/NSIS/INSTALL?view=3Dmarkup 6. http://prdownloads.sourceforge.net/mingw/MinGW-3.1.0-1.exe?download -- John |