[1.33.1][iostreams] Timetable and patches
Hi All, Do we have any timetable for 1.33.1? Will Bryant just pointed out a memory leak in Iostreams, and I'd like to post an Iostreams patch if 1.33.1 will not be ready reasonably soon. One more thing: before there was talk of 1.33.1, I prepared a patch for 1.33 to fix some minor stuff, but had trouble generating a single patch that worked for all the sourceforge 1.33 archives. Should it be necessary to post several patches, or was I doing something wrong? -- Jonathan Turkanis www.kangaroologic.com
On Sep 24, 2005, at 2:31 PM, Jonathan Turkanis wrote:
Hi All,
Do we have any timetable for 1.33.1?
I wrote on the main Boost web page that we were hoping to have a beta ready by September 30th... I'd still like to, but we have a lot of failures to address, still.
Will Bryant just pointed out a memory leak in Iostreams, and I'd like to post an Iostreams patch if 1.33.1 will not be ready reasonably soon.
I consider this serious enough that it should go into 1.33.1.
One more thing: before there was talk of 1.33.1, I prepared a patch for 1.33 to fix some minor stuff, but had trouble generating a single patch that worked for all the sourceforge 1.33 archives.
Should it be necessary to post several patches, or was I doing something wrong?
That's odd, since they should all be the same. Was it a problem with line endings, perhaps? Doug
Douglas Gregor wrote:
On Sep 24, 2005, at 2:31 PM, Jonathan Turkanis wrote:
Hi All,
Do we have any timetable for 1.33.1?
I wrote on the main Boost web page that we were hoping to have a beta ready by September 30th... I'd still like to, but we have a lot of failures to address, still.
Will Bryant just pointed out a memory leak in Iostreams, and I'd like to post an Iostreams patch if 1.33.1 will not be ready reasonably soon.
I consider this serious enough that it should go into 1.33.1.
One more thing: before there was talk of 1.33.1, I prepared a patch for 1.33 to fix some minor stuff, but had trouble generating a single patch that worked for all the sourceforge 1.33 archives.
Should it be necessary to post several patches, or was I doing something wrong?
That's odd, since they should all be the same. Was it a problem with line endings, perhaps?
That's what it seems to be. For example, the file boost/detail/utf8_codecvt_facet.hpp has unix line-endings in the tar.gz archive, but DOS line-endings in the zip archive. I guess this happened for some iostreams files, too. The .tar.gz archive is a few k smaller than the .zip archive.
Doug
-- Jonathan Turkanis www.kangaroologic.com
Pavel Vozenilek wrote:
"Jonathan Turkanis" wrote:
For example, the file boost/detail/utf8_codecvt_facet.hpp has unix line-endings in the tar.gz archive, but DOS line-endings in the zip archive.
That's an expected behaviour.
Why would it be just a few files?
/Pavel
-- Jonathan Turkanis www.kangaroologic.com
On Sep 24, 2005, at 6:27 PM, Jonathan Turkanis wrote:
Douglas Gregor wrote:
That's odd, since they should all be the same. Was it a problem with line endings, perhaps?
That's what it seems to be. For example, the file boost/detail/utf8_codecvt_facet.hpp has unix line-endings in the tar.gz archive, but DOS line-endings in the zip archive. I guess this happened for some iostreams files, too. The .tar.gz archive is a few k smaller than the .zip archive.
Okay, that part is intentional. We always provide Unix line endings for tarballs and DOS line endings for ZIP files, so I guess you will have to provide two patches... but you can probably just convert the line endings from one to create the second. Doug
Douglas Gregor wrote:
On Sep 24, 2005, at 6:27 PM, Jonathan Turkanis wrote:
Douglas Gregor wrote:
That's what it seems to be. For example, the file boost/detail/utf8_codecvt_facet.hpp has unix line-endings in the tar.gz archive, but DOS line-endings in the zip archive. I guess this happened for some iostreams files, too. The .tar.gz archive is a few k smaller than the .zip archive.
Okay, that part is intentional. We always provide Unix line endings for tarballs and DOS line endings for ZIP files, so I guess you will have to provide two patches... but you can probably just convert the line endings from one to create the second.
A quick sampling of the tar.gz archive seems to show that most files have DOS line endings, but that a few have UNIX line-endings. -- Jonathan Turkanis www.kangaroologic.com
On Sep 25, 2005, at 11:19 AM, Jonathan Turkanis wrote:
Douglas Gregor wrote:
On Sep 24, 2005, at 6:27 PM, Jonathan Turkanis wrote:
Douglas Gregor wrote:
That's what it seems to be. For example, the file boost/detail/utf8_codecvt_facet.hpp has unix line-endings in the tar.gz archive, but DOS line-endings in the zip archive. I guess this happened for some iostreams files, too. The .tar.gz archive is a few k smaller than the .zip archive.
Okay, that part is intentional. We always provide Unix line endings for tarballs and DOS line endings for ZIP files, so I guess you will have to provide two patches... but you can probably just convert the line endings from one to create the second.
A quick sampling of the tar.gz archive seems to show that most files have DOS line endings, but that a few have UNIX line-endings.
That's... unnerving. I'll look into it. Doug
participants (10)
-
Douglas Gregor -
Jonathan Turkanis -
Pavel Vozenilek