When LINQ translates the below syntax to SQL, the (inner) where clause gets moved to the outer-most query. That's super-unfriendly to the database. I wrote this like Hibernate's HQL (is this appropriate?), and I've written SQL for many moons.
Can anyone help explain what gives, or point me in the way of a resolution?
var rc = (
from dv in (
from dv_j in (
from x in adc.JobManagement
join j in adc.Job on x.JobId equals j.JobId
join js in adc.JobStatus on j.StatusId equals js.JobStatusId
join cm in adc.ClientManagement on j.ClientId equals cm.ClientId
join o in adc.User on cm.UserId equals o.UserId
join jm in adc.JobManagement on j.JobId equals jm.JobId
where
(x.UserId == aid || cm.UserId == aid)
&& (j.StatusDate == null || j.StatusDate >= getFromDate())
&& (jm.ManagementRoleCode == MR_MANAGER)
select new
{
j.JobId,
Job = j.InternalName == null ? j.ExternalName : j.InternalName,
JobStatusDate = j.StatusDate,
JobStatus = js.Code,
Owner = o.Username,
Role = jm.ManagementRoleCode
})
join s in adc.Submission on dv_j.JobId equals s.JobId into dv_s
from s in dv_s.DefaultIfEmpty()
select new
{
dv_j.JobId,
dv_j.Job,
dv_j.JobStatusDate,
dv_j.JobStatus,
dv_j.Owner,
dv_j.Role,
s.SubmissionId,
s.CandidateId,
s.SubmissionDate,
StatusDate = s.StatusDate,
StatusId = s.StatusId
})
join c in adc.Candidate on dv.CandidateId equals c.CandidateId into dv_c
join ss in adc.SubmissionStatus on dv.StatusId equals ss.SubmissionStatusId into dv_ss
from c in dv_c.DefaultIfEmpty()
from ss in dv_ss.DefaultIfEmpty()
orderby
dv.StatusId == null ? dv.StatusDate : dv.JobStatusDate descending,
dv.Job,
c.LastName,
c.NickName,
c.FirstName
select new Projects
{
Id = dv.JobId,
Project = dv.Job,
Submitted = dv.SubmissionDate,
Candidate = Form开发者_Go百科atIndividual(c.LastName, c.FirstName, c.NickName),
Status = dv.StatusId == null ? ss.Code : dv.JobStatus,
StatusDate = dv.StatusId == null ? dv.StatusDate : dv.JobStatusDate,
Role = dv.Role,
Owner = dv.Owner
});
Try breaking down the one statement into two. This would work as it cannot move the where to a place that doesn't exist yet. This does make multiple round trips to the database, but it is better than having most of several large tables being joined then culled. I would try this:
var inMemoryTable = (
from x in adc.JobManagement
join j in adc.Job on x.JobId equals j.JobId
join js in adc.JobStatus on j.StatusId equals js.JobStatusId
join cm in adc.ClientManagement on j.ClientId equals cm.ClientId
join o in adc.User on cm.UserId equals o.UserId
join jm in adc.JobManagement on j.JobId equals jm.JobId
where
(x.UserId == aid || cm.UserId == aid)
&& (j.StatusDate == null || j.StatusDate >= getFromDate())
&& (jm.ManagementRoleCode == MR_MANAGER)
select new
{
j.JobId,
Job = j.InternalName == null ? j.ExternalName : j.InternalName,
JobStatusDate = j.StatusDate,
JobStatus = js.Code,
Owner = o.Username,
Role = jm.ManagementRoleCode
});
var rc = (
from dv in (
from dv_j in inMemoryTable
join s in adc.Submission on dv_j.JobId equals s.JobId into dv_s
from s in dv_s.DefaultIfEmpty()
select new
{
dv_j.JobId,
dv_j.Job,
dv_j.JobStatusDate,
dv_j.JobStatus,
dv_j.Owner,
dv_j.Role,
s.SubmissionId,
s.CandidateId,
s.SubmissionDate,
StatusDate = s.StatusDate,
StatusId = s.StatusId
})
join c in adc.Candidate on dv.CandidateId equals c.CandidateId into dv_c
join ss in adc.SubmissionStatus on dv.StatusId equals ss.SubmissionStatusId into dv_ss
from c in dv_c.DefaultIfEmpty()
from ss in dv_ss.DefaultIfEmpty()
orderby
dv.StatusId == null ? dv.StatusDate : dv.JobStatusDate descending,
dv.Job,
c.LastName,
c.NickName,
c.FirstName
select new Projects
{
Id = dv.JobId,
Project = dv.Job,
Submitted = dv.SubmissionDate,
Candidate = FormatIndividual(c.LastName, c.FirstName, c.NickName),
Status = dv.StatusId == null ? ss.Code : dv.JobStatus,
StatusDate = dv.StatusId == null ? dv.StatusDate : dv.JobStatusDate,
Role = dv.Role,
Owner = dv.Owner
});
精彩评论