This is the mail archive of the cygwin mailing list for the Cygwin 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: "Couldn't allocate heap" - tried rebasing

On 11/22/2011 10:24 PM, Larry Hall (Cygwin) wrote:
On 11/22/2011 8:08 PM, Jon Clugston wrote:

Actually, I just noticed this remark:

"In summary, current Windows implementations make it
impossible to implement a perfectly reliable fork, and occasional
fork failures are inevitable."

in winsup/doc/overview2.sgml in the source tree. Does that mean that,
with the improvements mentioned above, we cannot expect important Cygwin
apps/scripts to always work reliably in a post-WinXP world? My
company has
been moving from Win2K/XP to Win7, so this would be important info
for us.

So how serious is the above remark? I don't see anything quite that
strongly-phrased in the FAQ. Maybe it should be mentioned there?

I would assume that "current Windows implementations" means XP and above. I have found it to be quite stable on Windows 7 once a rebase is done. I also believe that the possibility of "fork" failing has always been there - even in Cygwin 1.5. So, maybe the remark is not quite as scary as it might at first appear.

The fork issue is nothing new. It has existed for a long time. The 1.5 series was certainly not immune. The fact that fork failures may be more prevalent now than before has as much to do with the growth in the number of packages available with Cygwin as it does with the changes in the Windows environment that work against the fork implementation. And there have been efforts to combat the negative impacts of both of these changes, particularly in post 1.7.9 snapshots (and eventually packages. :-) ) My recommendation is to not worry about fork failures until you see them and then install the rebase package, read the readme, and follow the directions found there. In other words, don't worry more now than you did before. ;-)

OK. From that document I had gotten the impression that the inevitability of the problem was related to Address Space Layout Randomization introduced in Windows Vista, and therefore, XP had been in a better situation. If the probability of problems after the latest snapshot on Win7 is the same as it's always been on WinXP, then I guess we'll be no more or less likely than before to see these issues. Thanks.

| Jesse Ziser, Code Warrior |
| Applied Research Labs: UT |

Problem reports:
Unsubscribe info:

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