This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree-ssa] Merge status

On Wednesday, March 17, 2004, at 06:31 AM, wrote:
Assuming that's the case, I would _expect_ that these developers will happily
fix their code when the problems are pointed out.

I've had that experience to some degree, but also the opposite. People that don't want to maintain code that just want it to compile, because it used to compile before. They really hate it when it just _stops_ compiling. The code is often not theirs, but rather comes from a 3rd party or they are doing porting (they are the 3rd party). They just want it to compile.

Additionally, there is the case where you want to develop and test a new compiler against the entire OS build, but you can't because things don't compile, and therefore, you can't test, therefore you loose testing, which is bad. The project owners are busy working on real development and don't always want to be bothered with the minor maintenance. They'll get around to doing it, but they want to schedule it and do it at a different time. The problem is, in the intervening timeframe you've lost all testing that could have been done.

Also, even if they were willing to do the changes, they want to do them in their top of tree, which, isn't the stable testing source you wanted to test, but rather the build of the day software of the next OS that may or may not work, which defeats the entire purpose of building and testing it.

These issues aren't new or surprising, but I thought I would point them out to balance your statement.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]