[jira] Updated: (GERONIMO-2863) Mail 1.4 InternetHeaders behavior difference compared to Sun's
[ https://issues.apache.org/jira/browse/GERONIMO-2863?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Warner updated GERONIMO-2863: --- Attachment: Geronimo-2863_Specs.patch The test failures was due to a logic error in my code. I found it and fixed it. The new code pass all the tests on my box. > Mail 1.4 InternetHeaders behavior difference compared to Sun's > -- > > Key: GERONIMO-2863 > URL: https://issues.apache.org/jira/browse/GERONIMO-2863 > Project: Geronimo > Issue Type: Bug > Security Level: public(Regular issues) > Components: specs >Reporter: Daniel Kulp > Assigned To: Jason Warner >Priority: Minor > Attachments: Geronimo-2863_Specs.patch > > > The InternetHeaders object's parsing is less "forgiving" of headers that > don't meet specs compared to Sun's. > Sun's implementation allows headers to be using "unix style" (just LF) and > will parse them fine. > The 1.4 version of Geronimo's only breaks on the CR and then blindly consumes > the next char (should be the LF). If the stream only has LF's, everything > is read into a single header. > I know the spec says the stream should be CR/LF so the behavior is fine for > proper streams. However, it is a behavior change that COULD make it not a > drop in replacement for Suns. (in our case, we had some test data that were > "svn:eol_style: native" that passed on Windows, but not on Unix. Changing > them to "svn:eol_style: CRLF" fixed it.) -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (GERONIMO-2863) Mail 1.4 InternetHeaders behavior difference compared to Sun's
[ https://issues.apache.org/jira/browse/GERONIMO-2863?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Warner updated GERONIMO-2863: --- Attachment: (was: Geronimo-2863_Specs.patch) > Mail 1.4 InternetHeaders behavior difference compared to Sun's > -- > > Key: GERONIMO-2863 > URL: https://issues.apache.org/jira/browse/GERONIMO-2863 > Project: Geronimo > Issue Type: Bug > Security Level: public(Regular issues) > Components: specs >Reporter: Daniel Kulp > Assigned To: Jason Warner >Priority: Minor > > The InternetHeaders object's parsing is less "forgiving" of headers that > don't meet specs compared to Sun's. > Sun's implementation allows headers to be using "unix style" (just LF) and > will parse them fine. > The 1.4 version of Geronimo's only breaks on the CR and then blindly consumes > the next char (should be the LF). If the stream only has LF's, everything > is read into a single header. > I know the spec says the stream should be CR/LF so the behavior is fine for > proper streams. However, it is a behavior change that COULD make it not a > drop in replacement for Suns. (in our case, we had some test data that were > "svn:eol_style: native" that passed on Windows, but not on Unix. Changing > them to "svn:eol_style: CRLF" fixed it.) -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (GERONIMO-2863) Mail 1.4 InternetHeaders behavior difference compared to Sun's
[ https://issues.apache.org/jira/browse/GERONIMO-2863?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Warner updated GERONIMO-2863: --- Attachment: Geronimo-2863_Specs.patch The InternetHeaders class has been modified so that when it is loading a new header a lone LF character will be treated the same as the CRLF combination. > Mail 1.4 InternetHeaders behavior difference compared to Sun's > -- > > Key: GERONIMO-2863 > URL: https://issues.apache.org/jira/browse/GERONIMO-2863 > Project: Geronimo > Issue Type: Bug > Security Level: public(Regular issues) > Components: specs >Reporter: Daniel Kulp > Assigned To: Jason Warner >Priority: Minor > Attachments: Geronimo-2863_Specs.patch > > > The InternetHeaders object's parsing is less "forgiving" of headers that > don't meet specs compared to Sun's. > Sun's implementation allows headers to be using "unix style" (just LF) and > will parse them fine. > The 1.4 version of Geronimo's only breaks on the CR and then blindly consumes > the next char (should be the LF). If the stream only has LF's, everything > is read into a single header. > I know the spec says the stream should be CR/LF so the behavior is fine for > proper streams. However, it is a behavior change that COULD make it not a > drop in replacement for Suns. (in our case, we had some test data that were > "svn:eol_style: native" that passed on Windows, but not on Unix. Changing > them to "svn:eol_style: CRLF" fixed it.) -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
