Re: URGENT: Need to build new version based on 4.1.4
Hi Jim, I saw that you updated the metadata for 4.1.5. One file is still missing, see: https://bz.apache.org/ooo/show_bug.cgi?id=127552#c2 Regards, Matthias Am 18.11.2017 um 22:20 schrieb Jim Jagielski: > We do have the AOO415 branch! :) > >> On Nov 18, 2017, at 2:11 PM, Patricia Shanahanwrote: >> >> There is a bug in my LinkManager changes. Specifically, I made the totally >> newbie mistake of accessing memory based on a pointer without allowing for >> it being null, and it turns out it can be null :-( >> >> The primary report for this is >> https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after >> copying all content". >> >> I would have expected the failure to be some form of access violation. >> However, recent e-mails have convinced me that some paths include catch >> blocks that conceal all exceptions, so it may be the root cause of other >> problems. >> >> I have checked a fix into the trunk, revision 1814994. However, trunk has >> other changes. >> >> The simplest way to check whether other bug reports are due to this is to >> build a version that is identical to 4.1.4 except for the one fix. Anything >> that works with that but not with 4.1.4 is a duplicate. Of course, the more >> duplicates we have the worse the problem and the more urgent a fix. >> >> I am going ahead and doing the fix in my 4.1.5 working copy, but activating >> 4.1.5 is a broader decision. >> >> >> - >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >> For additional commands, e-mail: dev-h...@openoffice.apache.org >> > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > > smime.p7s Description: S/MIME Cryptographic Signature
Re: URGENT: Need to build new version based on 4.1.4
Now that I have a bit more time, I wanted to thank Jim for raising the question. I really want people to review my code and question anything that seems dubious. On 11/19/2017 7:21 AM, Patricia Shanahan wrote: The null check inside the mAutoAskUpdateAllLinks test is something of an over-reaction on my part, in the direction of defensive programming. mAutoAskUpdateAllLinks is only true in two specific cases, both of which I have reviewed. The bug was doing the parent look-up outside that if block. I have tested forcing the parent to null in a debugger. It does not crash. I suspect it would not necessarily pick the right parent window if multiple windows are open. On 11/19/2017 7:11 AM, Jim Jagielski wrote: Just to be clear: SetUserAllowsLinkUpdate(pLink, GetUserAllowsLinkUpdate(parent)); is OK if parent == NULL? That is, does GetUserAllowsLinkUpdate() work OK if passed NULL? Or, if parent is NULL, should we skip calling SetUserAllowsLinkUpdate()? On Nov 18, 2017, at 3:49 PM, Patricia Shanahanwrote: I'm attaching the patch in case anyone else wants to build their own test version. On 11/18/2017 11:11 AM, Patricia Shanahan wrote: There is a bug in my LinkManager changes. Specifically, I made the totally newbie mistake of accessing memory based on a pointer without allowing for it being null, and it turns out it can be null :-( The primary report for this is https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after copying all content". I would have expected the failure to be some form of access violation. However, recent e-mails have convinced me that some paths include catch blocks that conceal all exceptions, so it may be the root cause of other problems. I have checked a fix into the trunk, revision 1814994. However, trunk has other changes. The simplest way to check whether other bug reports are due to this is to build a version that is identical to 4.1.4 except for the one fix. Anything that works with that but not with 4.1.4 is a duplicate. Of course, the more duplicates we have the worse the problem and the more urgent a fix. I am going ahead and doing the fix in my 4.1.5 working copy, but activating 4.1.5 is a broader decision. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
On Sun, Nov 19, 2017 at 8:08 AM, Jim Jagielskiwrote: > I think I will spend some time creating some 4.1.5-dev builds for people to > play with... > > > On Nov 19, 2017, at 10:40 AM, Patricia Shanahan wrote: > > > > Jim, > That would be great. :) > > > I am going to be away from computers for most of the day. If you have > everything you want tested in 4.1.5 checked in, I would like you to trigger > a request for general testing. > > > > In particular, anyone who is working a bug report that is a regression > between 4.1.3 and 4.1.4 should test it with the 4.1.5 branch, regardless of > the symptoms. > > > > Originally, I expected my null point bug to manifest as an access > violation, as is the case for the original report. I was monitoring > bugzilla for any similar reports. From dev@ discussion, I have learned > there are instances of catching and hiding all exceptions. That means data > structures could be left in an inconsistent state without AOO crashing or > the access violation being reported, and almost any symptom could result. > > > > On 11/19/2017 7:21 AM, Patricia Shanahan wrote: > >> The null check inside the mAutoAskUpdateAllLinks test is something of > an over-reaction on my part, in the direction of defensive programming. > mAutoAskUpdateAllLinks is only true in two specific cases, both of which I > have reviewed. The bug was doing the parent look-up outside that if block. > >> I have tested forcing the parent to null in a debugger. It does not > crash. I suspect it would not necessarily pick the right parent window if > multiple windows are open. > >> On 11/19/2017 7:11 AM, Jim Jagielski wrote: > >>> Just to be clear: > >>> > >>> SetUserAllowsLinkUpdate(pLink, GetUserAllowsLinkUpdate(parent)); > >>> > >>> is OK if parent == NULL? That is, does GetUserAllowsLinkUpdate() > >>> work OK if passed NULL? Or, if parent is NULL, should we skip > >>> calling SetUserAllowsLinkUpdate()? > >>> > On Nov 18, 2017, at 3:49 PM, Patricia Shanahan wrote: > > I'm attaching the patch in case anyone else wants to build their own > test version. > > On 11/18/2017 11:11 AM, Patricia Shanahan wrote: > > There is a bug in my LinkManager changes. Specifically, I made the > totally newbie mistake of accessing memory based on a pointer without > allowing for it being null, and it turns out it can be null :-( > > The primary report for this is https://bz.apache.org/ooo/ > show_bug.cgi?id=127581, "Writer crashes after copying all content". > > I would have expected the failure to be some form of access > violation. However, recent e-mails have convinced me that some paths > include catch blocks that conceal all exceptions, so it may be the root > cause of other problems. > > I have checked a fix into the trunk, revision 1814994. However, > trunk has other changes. > > The simplest way to check whether other bug reports are due to this > is to build a version that is identical to 4.1.4 except for the one fix. > Anything that works with that but not with 4.1.4 is a duplicate. Of course, > the more duplicates we have the worse the problem and the more urgent a fix. > > I am going ahead and doing the fix in my 4.1.5 working copy, but > activating 4.1.5 is a broader decision. > > > - > > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > > For additional commands, e-mail: dev-h...@openoffice.apache.org > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > >>> > >>> > >>> - > >>> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > >>> For additional commands, e-mail: dev-h...@openoffice.apache.org > >>> > >> - > >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > >> For additional commands, e-mail: dev-h...@openoffice.apache.org > > > > - > > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > > For additional commands, e-mail: dev-h...@openoffice.apache.org > > > > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > > -- -- MzK "Only the truth will save you now." -- Ensei Tankado, "Digital Fortress"
Re: URGENT: Need to build new version based on 4.1.4
I think I will spend some time creating some 4.1.5-dev builds for people to play with... > On Nov 19, 2017, at 10:40 AM, Patricia Shanahanwrote: > > Jim, > > I am going to be away from computers for most of the day. If you have > everything you want tested in 4.1.5 checked in, I would like you to trigger a > request for general testing. > > In particular, anyone who is working a bug report that is a regression > between 4.1.3 and 4.1.4 should test it with the 4.1.5 branch, regardless of > the symptoms. > > Originally, I expected my null point bug to manifest as an access violation, > as is the case for the original report. I was monitoring bugzilla for any > similar reports. From dev@ discussion, I have learned there are instances of > catching and hiding all exceptions. That means data structures could be left > in an inconsistent state without AOO crashing or the access violation being > reported, and almost any symptom could result. > > On 11/19/2017 7:21 AM, Patricia Shanahan wrote: >> The null check inside the mAutoAskUpdateAllLinks test is something of an >> over-reaction on my part, in the direction of defensive programming. >> mAutoAskUpdateAllLinks is only true in two specific cases, both of which I >> have reviewed. The bug was doing the parent look-up outside that if block. >> I have tested forcing the parent to null in a debugger. It does not crash. I >> suspect it would not necessarily pick the right parent window if multiple >> windows are open. >> On 11/19/2017 7:11 AM, Jim Jagielski wrote: >>> Just to be clear: >>> >>> SetUserAllowsLinkUpdate(pLink, GetUserAllowsLinkUpdate(parent)); >>> >>> is OK if parent == NULL? That is, does GetUserAllowsLinkUpdate() >>> work OK if passed NULL? Or, if parent is NULL, should we skip >>> calling SetUserAllowsLinkUpdate()? >>> On Nov 18, 2017, at 3:49 PM, Patricia Shanahan wrote: I'm attaching the patch in case anyone else wants to build their own test version. On 11/18/2017 11:11 AM, Patricia Shanahan wrote: > There is a bug in my LinkManager changes. Specifically, I made the > totally newbie mistake of accessing memory based on a pointer without > allowing for it being null, and it turns out it can be null :-( > The primary report for this is > https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after > copying all content". > I would have expected the failure to be some form of access violation. > However, recent e-mails have convinced me that some paths include catch > blocks that conceal all exceptions, so it may be the root cause of other > problems. > I have checked a fix into the trunk, revision 1814994. However, trunk has > other changes. > The simplest way to check whether other bug reports are due to this is to > build a version that is identical to 4.1.4 except for the one fix. > Anything that works with that but not with 4.1.4 is a duplicate. Of > course, the more duplicates we have the worse the problem and the more > urgent a fix. > I am going ahead and doing the fix in my 4.1.5 working copy, but > activating 4.1.5 is a broader decision. > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org >>> >>> >>> - >>> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >>> For additional commands, e-mail: dev-h...@openoffice.apache.org >>> >> - >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >> For additional commands, e-mail: dev-h...@openoffice.apache.org > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
Jim, I am going to be away from computers for most of the day. If you have everything you want tested in 4.1.5 checked in, I would like you to trigger a request for general testing. In particular, anyone who is working a bug report that is a regression between 4.1.3 and 4.1.4 should test it with the 4.1.5 branch, regardless of the symptoms. Originally, I expected my null point bug to manifest as an access violation, as is the case for the original report. I was monitoring bugzilla for any similar reports. From dev@ discussion, I have learned there are instances of catching and hiding all exceptions. That means data structures could be left in an inconsistent state without AOO crashing or the access violation being reported, and almost any symptom could result. On 11/19/2017 7:21 AM, Patricia Shanahan wrote: The null check inside the mAutoAskUpdateAllLinks test is something of an over-reaction on my part, in the direction of defensive programming. mAutoAskUpdateAllLinks is only true in two specific cases, both of which I have reviewed. The bug was doing the parent look-up outside that if block. I have tested forcing the parent to null in a debugger. It does not crash. I suspect it would not necessarily pick the right parent window if multiple windows are open. On 11/19/2017 7:11 AM, Jim Jagielski wrote: Just to be clear: SetUserAllowsLinkUpdate(pLink, GetUserAllowsLinkUpdate(parent)); is OK if parent == NULL? That is, does GetUserAllowsLinkUpdate() work OK if passed NULL? Or, if parent is NULL, should we skip calling SetUserAllowsLinkUpdate()? On Nov 18, 2017, at 3:49 PM, Patricia Shanahanwrote: I'm attaching the patch in case anyone else wants to build their own test version. On 11/18/2017 11:11 AM, Patricia Shanahan wrote: There is a bug in my LinkManager changes. Specifically, I made the totally newbie mistake of accessing memory based on a pointer without allowing for it being null, and it turns out it can be null :-( The primary report for this is https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after copying all content". I would have expected the failure to be some form of access violation. However, recent e-mails have convinced me that some paths include catch blocks that conceal all exceptions, so it may be the root cause of other problems. I have checked a fix into the trunk, revision 1814994. However, trunk has other changes. The simplest way to check whether other bug reports are due to this is to build a version that is identical to 4.1.4 except for the one fix. Anything that works with that but not with 4.1.4 is a duplicate. Of course, the more duplicates we have the worse the problem and the more urgent a fix. I am going ahead and doing the fix in my 4.1.5 working copy, but activating 4.1.5 is a broader decision. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
The null check inside the mAutoAskUpdateAllLinks test is something of an over-reaction on my part, in the direction of defensive programming. mAutoAskUpdateAllLinks is only true in two specific cases, both of which I have reviewed. The bug was doing the parent look-up outside that if block. I have tested forcing the parent to null in a debugger. It does not crash. I suspect it would not necessarily pick the right parent window if multiple windows are open. On 11/19/2017 7:11 AM, Jim Jagielski wrote: Just to be clear: SetUserAllowsLinkUpdate(pLink, GetUserAllowsLinkUpdate(parent)); is OK if parent == NULL? That is, does GetUserAllowsLinkUpdate() work OK if passed NULL? Or, if parent is NULL, should we skip calling SetUserAllowsLinkUpdate()? On Nov 18, 2017, at 3:49 PM, Patricia Shanahanwrote: I'm attaching the patch in case anyone else wants to build their own test version. On 11/18/2017 11:11 AM, Patricia Shanahan wrote: There is a bug in my LinkManager changes. Specifically, I made the totally newbie mistake of accessing memory based on a pointer without allowing for it being null, and it turns out it can be null :-( The primary report for this is https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after copying all content". I would have expected the failure to be some form of access violation. However, recent e-mails have convinced me that some paths include catch blocks that conceal all exceptions, so it may be the root cause of other problems. I have checked a fix into the trunk, revision 1814994. However, trunk has other changes. The simplest way to check whether other bug reports are due to this is to build a version that is identical to 4.1.4 except for the one fix. Anything that works with that but not with 4.1.4 is a duplicate. Of course, the more duplicates we have the worse the problem and the more urgent a fix. I am going ahead and doing the fix in my 4.1.5 working copy, but activating 4.1.5 is a broader decision. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
Just to be clear: SetUserAllowsLinkUpdate(pLink, GetUserAllowsLinkUpdate(parent)); is OK if parent == NULL? That is, does GetUserAllowsLinkUpdate() work OK if passed NULL? Or, if parent is NULL, should we skip calling SetUserAllowsLinkUpdate()? > On Nov 18, 2017, at 3:49 PM, Patricia Shanahanwrote: > > I'm attaching the patch in case anyone else wants to build their own test > version. > > On 11/18/2017 11:11 AM, Patricia Shanahan wrote: >> There is a bug in my LinkManager changes. Specifically, I made the totally >> newbie mistake of accessing memory based on a pointer without allowing for >> it being null, and it turns out it can be null :-( >> The primary report for this is >> https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after >> copying all content". >> I would have expected the failure to be some form of access violation. >> However, recent e-mails have convinced me that some paths include catch >> blocks that conceal all exceptions, so it may be the root cause of other >> problems. >> I have checked a fix into the trunk, revision 1814994. However, trunk has >> other changes. >> The simplest way to check whether other bug reports are due to this is to >> build a version that is identical to 4.1.4 except for the one fix. Anything >> that works with that but not with 4.1.4 is a duplicate. Of course, the more >> duplicates we have the worse the problem and the more urgent a fix. >> I am going ahead and doing the fix in my 4.1.5 working copy, but activating >> 4.1.5 is a broader decision. >> - >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >> For additional commands, e-mail: dev-h...@openoffice.apache.org > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
I think together is best. :) > On Nov 18, 2017, at 6:01 PM, Patricia Shanahanwrote: > > I see you have checked in some Mac building changes. Would it be better to > keep the two fixes separate, or encourage testing both of them together? > > On 11/18/2017 1:20 PM, Jim Jagielski wrote: >> We do have the AOO415 branch! :) >>> On Nov 18, 2017, at 2:11 PM, Patricia Shanahan wrote: >>> >>> There is a bug in my LinkManager changes. Specifically, I made the totally >>> newbie mistake of accessing memory based on a pointer without allowing for >>> it being null, and it turns out it can be null :-( >>> >>> The primary report for this is >>> https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after >>> copying all content". >>> >>> I would have expected the failure to be some form of access violation. >>> However, recent e-mails have convinced me that some paths include catch >>> blocks that conceal all exceptions, so it may be the root cause of other >>> problems. >>> >>> I have checked a fix into the trunk, revision 1814994. However, trunk has >>> other changes. >>> >>> The simplest way to check whether other bug reports are due to this is to >>> build a version that is identical to 4.1.4 except for the one fix. Anything >>> that works with that but not with 4.1.4 is a duplicate. Of course, the more >>> duplicates we have the worse the problem and the more urgent a fix. >>> >>> I am going ahead and doing the fix in my 4.1.5 working copy, but activating >>> 4.1.5 is a broader decision. >>> >>> >>> - >>> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >>> For additional commands, e-mail: dev-h...@openoffice.apache.org >>> >> - >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >> For additional commands, e-mail: dev-h...@openoffice.apache.org > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
+1 for testing together. If it doesn't work we can go into analysis. Am 19. November 2017 00:01:35 MEZ schrieb Patricia Shanahan: >I see you have checked in some Mac building changes. Would it be better > >to keep the two fixes separate, or encourage testing both of them >together? > >On 11/18/2017 1:20 PM, Jim Jagielski wrote: >> We do have the AOO415 branch! :) >> >>> On Nov 18, 2017, at 2:11 PM, Patricia Shanahan wrote: >>> >>> There is a bug in my LinkManager changes. Specifically, I made the >totally newbie mistake of accessing memory based on a pointer without >allowing for it being null, and it turns out it can be null :-( >>> >>> The primary report for this is >https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after >copying all content". >>> >>> I would have expected the failure to be some form of access >violation. However, recent e-mails have convinced me that some paths >include catch blocks that conceal all exceptions, so it may be the root >cause of other problems. >>> >>> I have checked a fix into the trunk, revision 1814994. However, >trunk has other changes. >>> >>> The simplest way to check whether other bug reports are due to this >is to build a version that is identical to 4.1.4 except for the one >fix. Anything that works with that but not with 4.1.4 is a duplicate. >Of course, the more duplicates we have the worse the problem and the >more urgent a fix. >>> >>> I am going ahead and doing the fix in my 4.1.5 working copy, but >activating 4.1.5 is a broader decision. >>> >>> >>> >- >>> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >>> For additional commands, e-mail: dev-h...@openoffice.apache.org >>> >> >> >> - >> To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >> For additional commands, e-mail: dev-h...@openoffice.apache.org >> > >- >To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org >For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
I see you have checked in some Mac building changes. Would it be better to keep the two fixes separate, or encourage testing both of them together? On 11/18/2017 1:20 PM, Jim Jagielski wrote: We do have the AOO415 branch! :) On Nov 18, 2017, at 2:11 PM, Patricia Shanahanwrote: There is a bug in my LinkManager changes. Specifically, I made the totally newbie mistake of accessing memory based on a pointer without allowing for it being null, and it turns out it can be null :-( The primary report for this is https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after copying all content". I would have expected the failure to be some form of access violation. However, recent e-mails have convinced me that some paths include catch blocks that conceal all exceptions, so it may be the root cause of other problems. I have checked a fix into the trunk, revision 1814994. However, trunk has other changes. The simplest way to check whether other bug reports are due to this is to build a version that is identical to 4.1.4 except for the one fix. Anything that works with that but not with 4.1.4 is a duplicate. Of course, the more duplicates we have the worse the problem and the more urgent a fix. I am going ahead and doing the fix in my 4.1.5 working copy, but activating 4.1.5 is a broader decision. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
We do have the AOO415 branch! :) > On Nov 18, 2017, at 2:11 PM, Patricia Shanahanwrote: > > There is a bug in my LinkManager changes. Specifically, I made the totally > newbie mistake of accessing memory based on a pointer without allowing for it > being null, and it turns out it can be null :-( > > The primary report for this is > https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after > copying all content". > > I would have expected the failure to be some form of access violation. > However, recent e-mails have convinced me that some paths include catch > blocks that conceal all exceptions, so it may be the root cause of other > problems. > > I have checked a fix into the trunk, revision 1814994. However, trunk has > other changes. > > The simplest way to check whether other bug reports are due to this is to > build a version that is identical to 4.1.4 except for the one fix. Anything > that works with that but not with 4.1.4 is a duplicate. Of course, the more > duplicates we have the worse the problem and the more urgent a fix. > > I am going ahead and doing the fix in my 4.1.5 working copy, but activating > 4.1.5 is a broader decision. > > > - > To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org > For additional commands, e-mail: dev-h...@openoffice.apache.org > - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
Re: URGENT: Need to build new version based on 4.1.4
I'm attaching the patch in case anyone else wants to build their own test version. On 11/18/2017 11:11 AM, Patricia Shanahan wrote: There is a bug in my LinkManager changes. Specifically, I made the totally newbie mistake of accessing memory based on a pointer without allowing for it being null, and it turns out it can be null :-( The primary report for this is https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after copying all content". I would have expected the failure to be some form of access violation. However, recent e-mails have convinced me that some paths include catch blocks that conceal all exceptions, so it may be the root cause of other problems. I have checked a fix into the trunk, revision 1814994. However, trunk has other changes. The simplest way to check whether other bug reports are due to this is to build a version that is identical to 4.1.4 except for the one fix. Anything that works with that but not with 4.1.4 is a duplicate. Of course, the more duplicates we have the worse the problem and the more urgent a fix. I am going ahead and doing the fix in my 4.1.5 working copy, but activating 4.1.5 is a broader decision. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org Index: main/sfx2/source/appl/linkmgr2.cxx === --- main/sfx2/source/appl/linkmgr2.cxx (revision 1813359) +++ main/sfx2/source/appl/linkmgr2.cxx (working copy) @@ -158,9 +158,13 @@ SvBaseLinkRef* pTmp = new SvBaseLinkRef( pLink ); pLink->SetLinkManager( this ); aLinkTbl.Insert( pTmp, aLinkTbl.Count() ); - Window *parent = GetPersist()->GetDialogParent(); if (mAutoAskUpdateAllLinks) { + Window *parent = NULL; + SfxObjectShell* persist = GetPersist(); + if (persist != NULL) + parent = GetPersist()->GetDialogParent(); + SetUserAllowsLinkUpdate(pLink, GetUserAllowsLinkUpdate(parent)); } - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
URGENT: Need to build new version based on 4.1.4
There is a bug in my LinkManager changes. Specifically, I made the totally newbie mistake of accessing memory based on a pointer without allowing for it being null, and it turns out it can be null :-( The primary report for this is https://bz.apache.org/ooo/show_bug.cgi?id=127581, "Writer crashes after copying all content". I would have expected the failure to be some form of access violation. However, recent e-mails have convinced me that some paths include catch blocks that conceal all exceptions, so it may be the root cause of other problems. I have checked a fix into the trunk, revision 1814994. However, trunk has other changes. The simplest way to check whether other bug reports are due to this is to build a version that is identical to 4.1.4 except for the one fix. Anything that works with that but not with 4.1.4 is a duplicate. Of course, the more duplicates we have the worse the problem and the more urgent a fix. I am going ahead and doing the fix in my 4.1.5 working copy, but activating 4.1.5 is a broader decision. - To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org